Can anybody say to me how to introduce new code in ground using parallel phoenics????

Parallel PHOENICS operates on a user-specified number of subdomains. These are created automatically by splitting the total solution domain. Each subdomain is assigned to a different parallel processor, and it is enlarged at each edge for which there is a neighbouring subdomain by the addition of two layers of cells. These additional 'overlap' cells are used to store field values from the adjacent subdomains so as to facilitate data exchange between subdomains during the parallel computation. Each processor works only on its own subdomain - and using its own cell numbering system, starting from 1 in each direction.

GROUND coding that only uses current cell values should require no modification to work in parallel mode.

Difficulties may occur where, for instance,

- summations are made over cells that are not in the same subdomain; - references are made to non-neighbour variables that lie (or may lie) in another subdomain; and - references are made to non-neighbour cells by indices rather than physical coordinates.

In the first two cases it is necessary to make use of parallel utility routines to construct the necessary sums, or to exchange values between processors. These are described in the FAQ section of the User Support area on the CHAM web site (available to supported users in possession of a valid maintenance contract).

In the third case it is necessary to change the coding so that the cell indices are derived from the physical coordinates using the available grid information: indices based on the total grid are invalid, as a processor only knows about its own subdomain. This is good practice anyway: even for conventional sequential operation the 'hard-wired' use of cell indices means that any change to the grid will require a corresponding change to the GROUND coding.

More help is available to maintained users in possession of a valid maintenance contract by contacting support@cham.co.uk.