----------------------------------------------------------------- | CHAM | 1 | PHOENICS developments | |developments | ---- | at CHAM during 1992/1993 | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Major themes | | * Parallelization | | * Grid-restructuring (multi-blocking) | | * EXPERT | | * Incorporation of other solvers | | * Interfacing to other packages | | | | * Special items (a selection) | | * Non-staggered grid option (ie co-located velocities) | | * Chemical-kinetics add-on | | * Surface-to-surface radiation | | * Chemical vapour deposition | | | | * Improvements of existing features | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 2 | | |developments | ---- | Parallelization | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | Objectives | | 1. To create a parallelized version of PHOENICS for Parsytech| | multi-transputer machines | | | | 2. To create a parallelized version of PHOENICS for the CAMT | | "distributed-array processor" | | | | 3. To perform both tasks in a generic way, in order that | | porting to other MIMD and SIMD machines is easy | | | | Methods | | 1. CHAM leads the EEC-supported PASHA project, partnered by | | Parsytech, CAMT, EMIT, BIBA, and VUB | | 2. CHAM sponsors generic work at Southampton University | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 3 | | |developments | ---- | Parallelization (continued) | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Expected completion dates | | * The PASHA project is due for completion in September 1994 | | * Parallelized versions of PHOENICS will have been created | | and tested for both Parsytech and CAMT machines by Sept '93| | * Porting to other MIMD and SIMD machines will be performed | | when commercial orders are imminent | | | | * Remarks | | * By way of practice in parallelization, the KIVA (IC-engine)| | code is also being parallelized for both types of machine | | | | * The MIMD parallelization makes use of domain-decomposition | | as illustrated in "active demos" (see grid-restructuring) | | | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 4 | | |developments | ---- | Grid restructuring | | 1992-1993 | 16 | (multi-blocking) | |_____________|_______|___________________________________________| | The LINK scheme | | * Already available in PHOENICS is a cut-and-paste feature | | enabling (say) the grid A + B to be rearranged as shown: | | | | ---------------------------------- << this | | | | | | | | A | B | -------------- | | | | | | | | | ---------------------------------- | B | | | | | | | -------------------------- | | | | | | becomes this >> | A | | | | | | |_______________________________----------------__________________|
___________________________________________________________________ | CHAM | 5 | | |developments | ---- | Grid re-structuring (continued) | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Linking of the kind illustrated, if it involves one-to-one | | cell contact, can be achieved through PATCH and COVAL commands| | ie without GROUND coding | | | | * One-to-many cell contact, as in fine-grid-embedding, does | | require GROUND coding, exemplified in GXRSTR.FOR | | | | * The development tasks are:- | | (1) To permit block locations and block-to-block links to be | | activated at the grid-generation stage by mouse movements | | and clicks | | (2) To make special GROUND coding unnecessary | | | | * Completion is planned for the end of 1993 | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 6 | | |developments | ---- | EXPERT, ie automatic selection | | 1992-1993 | 16 | of numerical inputs | |_____________|_______|___________________________________________| | * A primitive version of EXPERT is supplied with PHOENICS 1.6.6 | | | | * This version is able to "home-in on" the best over- or under- | | relaxation factor for a single solved-for variable | | | | * Development tasks for 1993 are:- | | | | (1) to apply, test and refine the single-variable version | | of EXPERT over a wide range of conditions | | (2) to develop versions which can detect which variable has | | most effect on convergence and to optimise its factors | | (3) to handle several variables at the same time | | (4) to extend EXPERT to the making of initial-set-up choices | | including those of domain-decomposition for parallelization |_________________________________________________________________|
___________________________________________________________________ | CHAM | 7 | | |developments | ---- | Incorporation of other solvers | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * "Incorporation" means replacing EARTH by a different equation-| | solving program having special features, for example:- | | * completely open Fortran source code; or | | * adaptation to simulation of a particular equipment type; or| | * familiarity to an already existing user comunity. | | | | * The arrangement is: | | | | -------------- ----------------- --------------- | | | | | | | | | | | Satellite |--->-| solver |->-| PHOTON | | | | | | | | | | | | | | | | | | | --------------- ----------------- --------------- | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 8 | | |developments | ---- | Incorporation of other solvers | | 1992-1993 | 16 | (continued) | |_____________|_______|___________________________________________| | * CHAM has already incorporated several other codes in the | | system, including:- | | * CORA, CHAM's pre-PHOENICS combustion-chamber code | | * LEARN, an instructional code from Erlangen University | | * KIVA 1 and KIVA 2, public-domain codes for IC engines | | * VVV, a series of codes from CHAM-MEI, Moscow | | | | * Solvers to be incorporated in 1993/4 include:- | | * A structured-grid code simulating external flow around | | aircraft and missiles; | | * An unstructured-grid code used for aircraft design, but | | adaptable also to other external-flow problems | | * (Subject to not-yet-completed negotiations) a hybrid | | structured-plus-unstructured code. | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 9 | | |developments | ---- | Interfacing to other packages | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Interfaces already exist to: | | * PATRAN (but not yet PDA's latest issue) | | * IDEAS | | * FEMGEN/FEMVIEW | | * AVS | | * Data visualiser (Wavefront) | | * VVG (CHAM-MEI) | | * Interfaces to be created during 1993 include:- | | * CFView (Free University Brussels) | | * Fieldview (Intelligent Light) | | * Concept (Aries) | | | | * CHAM's policy is to provide or procure interfaces to all | | packages favoured by PHOENICS users. Suggestions are welcome. | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 10 | | |developments | ---- | Non-staggered-grid option | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | | | * Background | | * Some popular codes employ grids for which velocities are | | stored at cell centres, solving equations with variants of | | the Rhie-Chou scheme | | * The cartesian-direction velocities are solved for, even for | | polar or BFC grids | | * However, all known schemes produce non-physical solutions | | in some circumstances (eg "ripples" in velocity profiles, | | "chequer-board pressure fields,..)| | * Recent progress | | * CHAM-MEI has reviewed all schemes and invented a new one | | free from the above defects | | * Work for 1993 | | * The new scheme will be tested, then attached to PHOENICS | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 11 | | |developments | ---- | Chemical-kinetics upgrade | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Background | | * CHAM possesses, and still uses in-house, the CREK package | | created by Pratt and Wormeck in the 1970s. | | * An interface has been created, by Sher and Kravchuk in | | * Israel, to the public-domain CHEMKIN package. | | * The public-domain KIVA-2 program has a significant chemical- | kinetic capability. | | | | * Development tasks for 1993/4 | | * The capabilities, strengths and weaknesses of these packages | will be studied, during the course of funded projects | | * By adaptation, incorporation or new development, a flexible| | and user-friendly chemical-kinetic add-on module for | | PHOENICS will be created. | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 12 | | |developments | ---- | Surface-to-surface radiation | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Background | | * PHOENICS has the built-in six-flux model of radiation | | which is intended for absorbing and emitting media | | | | * This model can be used for fully transparent media, but | | gives results of uneven accuracy | | | | * CHAM is in the process of creating an interchange-factor | | add-on to PHOENICS, which will be capable of more accurate | | and efficient transparent-media simulations | | | | * Development tasks for 1993 | | * The work in progress will be completed and subjected to test | * A surface-to-surface module will then be finalised. | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 13 | | |developments | ---- | Chemical-vapour deposition | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * Background | | | | * CHAM is the coordinating partner in an EEC-supported | | project called ACCESS-CVD | | * The other parners are:- | | Siemens (Germany) | | ASM (Netherlands) | | Fraunhofer(Germany) | | Tech Univ Delft (Netherlands) | | | | * Planned outcome for end 1994 | | * A special-purpose version of PHOENICS, with customised | | menu-type front end and special GROUND coding, incorporating | appropriate physics and chemistry, will be created. | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 14 | | |developments | ---- | Improvements to existing features | | 1992-1993 | 15 | (priorities will be influenced by users) | |_____________|_______|___________________________________________| | * Reynolds-stress turbulence model | | * PHOENICS 1.6.6 does contain a Reynolds-stress model (see | | "active demos" | | * However, this is subject to certain (essentially needless) | | limitations (constant density, cartesian or polar grids) | | * These limitatations will be removed during 1993/4 | | | | * CLDA, QUICK and other difference schemes | | * Numerous schemes for reducing numerical diffusion have now | | been attached to PHOENICS, in open-Fortran form | | * They have been illustrated, however, only for scalars | | * Their equal applicability to the velocity variables will | | be demonstrated during 1993/4 | | * New schemes (eg SUCCA from Strathclyde) will be added. | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 15 | | |developments | ---- | Improvements of existing features | | 1992-1993 | 16 | (continued) | |_____________|_______|___________________________________________| | * The input menus and other features | | * The number of PHOENICS features which can be activated via | | the general menu will be increased significantly, including| | combustion and two-phase flow features | | * Grid-generation features will be improved, especially in | | regard to multi-blocking | | * PHOTON is being made menu-driven | | | | * BFC equation formulations | | * The existing BFC formulations for the mass and diffusion | | fluxes through cell walls generate asymmetries for highly | | non-orthogonal grids; they will be improved. | | * Some of the viscous shear and turbulence generation terms | | are also less accurate than they can be, and will be | | made better | |_________________________________________________________________|
___________________________________________________________________ | CHAM | 16 | | |developments | ---- | Concluding remarks | | 1992-1993 | 16 | | |_____________|_______|___________________________________________| | * User-support-generated tasks | | * CHAM's User-Support team is part of the Development Division | in order that developments should be stimulated by the | | requests for assistance by users. | | * Many improvements are made as a result of these requests. | | * This policy will continue. | | | | * Why CHAM needs to know what users really want | | * Some semi-complete developments, eg stress analysis in | | solids have been halted because of low user interest. | | * Other developments may not have started because users' | | needs have been incorrectly perceived. | | | | * CHAM welcomes users' suggestions, especially precise ones. | |_________________________________________________________________|