PHOENICS was first released as a commercial CFD code in London, in October 1981.

1
Message for the Tokyo PHOENICS-User Meetingfrom
Brian Spalding

• Welcome
• PHOENICS was first released as a commercial CFD
  code in London, in October 1981.

A year later it was introduced to Japan by CRC,
and Japanese Users of PHOENICS have been, since
then, among the most ingenious and productive in
the world.
I therefore welcome this opportunity to address
them and to explain in what ways, thirty years
after its birth, PHOENICS is still developing, in
response to the changing needs of its users.
2
Message for the Tokyo PHOENICS-User
MeetingSummary

• The needs of PHOENICS users are not all the same.
• In the past, CHAM has tried to satisfy all by
  means of a single data-input module, the
  VR-EDITOR, operating in graphical interactive
  mode.

• There are two drawbacks
• VRE presents too many choices for some users and
• it cannot activate all PHOENICS capabilities.

Now therefore CHAM has added new input
procedures which meet the differing needs of five
different classes of user.
These will now be described.
3
Recognising five different groups of users 1.
those needing Gateways
PHOENICS is a General-Purpose package but most
of its users have special purposes.

• User Group 1 wants easy-to-use simulation-and-desi
  gn software for particular processes, e.g.
• heat exchangers (shell-and-tube, plate,
  air-cooled )
• chemical reactors (CVD, stirred-tank,
  fluidised-bed...)
• environmental flows (wind-farm, city pollution,
  fires )
• pumps and fans (gear-wheel, bladed,
  piston-cylinder... )

To this group CHAM offers special-purpose
stand-alone packages, accessed through
application-specific Gateways.
4
Recognising five different groups of users
(continued)2. Gateway Creators

• CHAM has itself created several gateways and it
  will create any more which customers commission.
  But users of Group 2 are those who wish to
  create their own.
• To this group CHAM offers

• Examples of parameterised relational-data-input
  files
• explanations of these, and assistance, in the
  creation of new ones.

• PRELUDE, which automatically converts these into
  Gateway packages which exactly suit the needs of
  the defined Group-1 users.

5
Recognising five different groups of users
(continued)3. Those requiring new features

• Gateways enable users easily to access existing
  features of PHOENICS but Group-3 users wish
  involve features which do not yet exist, for
  example
• unusual geometries not fitting structured grids
• materials with unusual thermo-physical
  properties
• never-before-tried turbulence models or
• novel equation-solving algorithms.
• To this group CHAM offers

• In-Form which adds new features without new
  Fortran or C coding
• advice and assistance for those who request it
• a library of cases for copying.

6
Recognising five different groups of users
(continued)4. students and teachers

• PHOENICS can be used for learning about fluid
  mechanics, heat transfer, chemical reaction,
  solid mechanics and their engineering and
  environmental effects. For Group-4 members, i.e.
  students and teachers, CHAM offers

• PHOENICS-For-Teaching, which comprises
• an extremely-simple interface
• a teacher mode which creates cases to be run
• a student mode which facilitates running of
  cases
• automated calculation and display of results
• multi-language text capability
• an organisational framework which enables
  contributors to collaborate internationally.

7
Recognising five different groups of users
(continued)5. Already-experienced users

• Experienced users of PHOENICS may require no
  fundamental changes. As Group 5 they desire only
  that
• all familiar features should continue to exist
• changes should be only in greater speed of
  operation, richness of options, freedom from
  failures.
• To this group CHAM offers

• Continued development to ensure that the VR
  Editor, the Earth solver, the VR Viewer, PHOTON.
  POLIS, Parallel Operation, etcetera remain in use
  and are improved in detail.
• Free-of-charge support to maintained users.

8
Developments relating to Group 1An example the
hen-house gateway

• Let us suppose that the task is to predict air
  temperature, humidity and composition within a
  hen-battery building.

• It is a CFD problem so PHOENICS can be used to
  solve it. But how?

• Input data include the geometry (what is where)
  the production rates of heat and pollutant from
  each cage the air-supply rate and temperature.

• Free (gravity-induced) convection is as
  important as forced convection.

9
CFD for the hen-house Ease of use is essential

• Specialists in the animal-production-environment
  know little about CFD. They need programs devised
  especially for them.

• .

• Data input should be made easy, so as to save
  time and minimise mistakes.

• Simulation results should be presented in a way
  which makes them easy to inspect and interpret.

10
The CFD-Gateway conceptWhat a hen-house gateway
might look like
Picture of Gateway menu panel with part of a
battery of chicken- cages in its graphics window.
Button rows are standard for all gateways.
Clicking on items in the object tree reveals
what is specific to this gateway..
11
Developments relating to Group 1input data
which the user will supply

• The Gateway user will supply, via menu boxes, the
  data which he knows about e.g.
• size, number and arrangement of cages
• location, temperature and flow rate of air
  supply
• attributes of chickens (heat output, oxygen
  consumpton, air-pollution rate).

• The Gateway user will not be asked to choose
• turbulence models,
• grid type or fineness
• sweep number
• relaxation and other numerical parameters.

The Gateway creator (User-Group 2) will have
already built choices into the Gateway-design
script.
12
The CFD-Gateway conceptLooking at the results
After its user is satisfied with the input data,
clicking the run button launches the simulation
and displays results as velocity vectors (right)
or contour displays (left) on accordance with
pre-set orders, or the users menu-expressed
wishes.
The left-hand picture shows the SMEL (odour)
contours.
13
The CFD-Gateway conceptWhat the Group-1 user
would learn

• Printed numerical results of interest to the user
  might include
• locations and maximum values of temperature,
  air-pollutant, etcetera
• electric power and heat input needed for
  specified conditions.

By varying the input conditions and observing the
changes in results, the user would thus be
enabled to determine how to optimise hen-house
design and operation.
Thus the animal-production specialist could
become, without expensive training, a user of
PHOENICS and CFD.
14
Second user GroupThe Gateway Creators
Users of Group 1 are able to benefit from CFD
when suitable PHOENICS Gateways have been
prepared for them by someone else. By whom?
By Group-2 users, who have learned how to create
Gateways.
Managers please note Personnel for Groups 1 and
2 utilise different skills, and convey different
benefits. A well-balanced team requires persons
of both kinds.
15
Developments relating to User Group 2how to
create the hen-house gateway

• To create a new gateway, one need not start from
  scratch for the PRELUDE package exists, with
  ready-to-use general framework and customising
  tools.

Here is shown its standard top-menu bar.
These provide the frequently-required functions
of opening cases, introducing or deleting
objects, selecting them for attribute
modification, running simulation calculations,
etcetera.
Tutorials exist which explain these.
16
Developments relating to Group 2 Objects and
their attributes
Here is shown an object tree which might be
created for a hen-house gateway.
Clicking on an item enables its attributes to be
reviewed, and edited if need be.
These items control results-display items, viz
stream-lines and cutting planes for contours
and vectors and
these the in-and outflow of ventilating air.
Here are the stored and solved-for 3D variables.
Note that SMEL, of which chickens are a source,
is one of them.
17
Developments relating to Group 2Inserting an
array of cage objects
It would be would be tedious to introduce the
cages one by one so PRELUDE offers an array
facility seen below. Note that expressions can be
entered.
Here shed is the (highlighted) name of the
array. Cage is the name of the object arrayed.
18
Comparison ofVR-Editor and PRELUDE
The PHOENICS VR Editor facilitates the setting
up only of single-instance flow-simulation
scenarios.
For example, only numbers can be typed into its
menu boxes, thus but not expressions as in
PRELUDE
VRE can display scenarios created by PRELUDE.
But it lacks PRELUDEs array-specification
feature.
PRELUDE.is therefore the preferred user interface
for PHOENICS.
19
Developments relating to Group 2what the
Gateway-creator must do

• Just as VRE receives instructions from Q1 written
  in PHOENICS Input Language (i.e. PIL),
• PRELUDE receives instructions from Q3 written in
  PRELUDE Script Language (i.e. PSL).

Group-2 users can learn PSL, and so write Q3
files or they can express their needs in
parameterised Q1 (i.e. PIL) files. Then PRELUDE
can translate these into Q3 files. Therefore
Group-2 users need not learn PSL.
Tutorials explain how to write parameterised Q1s
and there exist input-file-library examples and a
power-point presentation.
20
Developments relating to Group 2data input
How PRELUDE, Q1, Q3, VRE, etc fit together is
shown below.
Gateway creators express choices of turbulence
models and numerical parameters via PIL in Q1.
They must also make wise choices about whether
(or when) to use Space-Averaged or
Detailed-Geometry CFD, both of which PHOENICS can
supply.
What are SACFD and DGCFD? See next slide.
21
Space-Averaged- and Detailed-Geometry CFD for
heat exchangers
SACFD
DGCFD
Tubes in a heat exchanger are too numerous for
DGCFD to be used for the whole equipment so
formulae for volumetric heat transfer are used,
derived from
results of DGCFD applied to parts of the tube
bundle.
The DGCFD results are obtained first. Then the
results are parameterised for use later in SACFD.
22
DG-CFD as the extrapolator from experimental data
Question But there exist experimental data for
volumetric heat-transfer and friction
coefficients of tube bundles, as shown on the
right. Why not use them?
Answer They should be used but there are never
enough of them and they seldom reveal the
effects of, say, property variations with
temperature. So they are best used for
calibrating DG-CFD simulation models.
23
SA- and DG-CFDapplied to the hen house
Combined SA and DG can be applied to hen-house
simulation simply by setting the array parameters
to 1,1,1, with the result on the right
Now all the computational cells can be
concentrated in the smaller domain surrounding
the single cage so transfers of heat, matter and
momentum from hen to air can be more accurately
simulated.
Doing so at various Reynolds numbers and wind
angles allows formulae to be developed which can
be used in whole-hen-house SA-CFD simulations.
24
Detailed-geometry CFDThe pecking chicken, 1.
PRELUDE can also be used for moving objects Here
is a chicken, with the detailed-geometry grid
set up for a time-dependent simulation.
25
Detailed-geometry CFDThe pecking chicken, 2.
PRELUDE enables the motion of the chicken to be
prescribed by way of formulae. Here is the
formula to be used for describing its periodic
pecking motion.
The PHOENICS MOFOR (i.e. MOving-Frame- Of
Reference) technique is activated.
26
Detailed-geometry CFDThe pecking chicken, 3.
Here is an animated view of some results of
simulation. Note the air-velocity vectors and
surface -pressures.
This shows (more for entertainment than
necessity) what can be done with
detailed-geometry CFD hen-house designers need
not go so far!
27
The Moving-Frame-of Reference Technique some
details
MOFOR consists of two features, namely (1)
description of the motion and (2) Creating the
response of the fluid to that motion.
The first uses a technique borrowed from the
motion-picture industry (Titanic, Babe) a
MOF file is generated, as shown on the next
slide. It dictates the translations, rotations
and corresponding velocities and accelerations of
torso and limbs.
The second exploits whatever source-and-sink
mechanism is present in the underlying CFD solver
(here, of course, PHOENICS).
28
The MOF file generated by PRELUDE for the pecking
hen
Below is the first part of the MOF. Below it are
the corresponding tables of numbers which
represent the coordinates of the body and limbs
at a succession of times.
q1object chookb OFFSET 0.000000
0.000000 0.043844 localrot chookb
JOINT chookb localaxis locrot0
OFFSET 0.000000 0.2050000
0.200000 CHANNELS 1 Xrotation
End Site End Site
Prelude generated MOF file using time dependent
degrees of freedomHIERARCHY UNITS METRES ROOT
Cham JOINT world groupobject
world OFFSET 0.000000 0.000000 0.000000
JOINT chicke groupobject chicke
OFFSET 0.500000 0.500000 0.000000
JOINT chookbcore
29
CFD for hen-house simulationsummary of the
argument.

• It has been argued, with exemplification, that
• Current CFD techniques are adequate for the
  simulation of flow of air and pollutants in
  hen-houses.
• Both space-averaged and detailed-geometry
  techniques are needed.
• Even moving-object scenarios can be simulated.

This example has shown how Group-2 users of
PHOENICS can create Gateways for Group-1 users.
Who may these Group-2 Users be?
CHAM, or CHAM-J, personnel of course
but also any user who is willing to learn PIL.
30
Other existing Gateways SHELLFLO
SHELLFLO assists designers of shell-and-tube heat
exchangers to understand the flow inside their
equipment.
Here the x, y and z locations of a baffle in a
shell-and-tube heat exchanger are being fixed so
as always to fit the shell.
Volumetric friction- and heat-transfer-
coefficient formulae, in terms of local Reynolds
and Prandt Number can be input in this manner.
31
Other existing Gateways The Virtual WindTunnel,
VWT 1
Velocity vectors around a sphere, in a structured
grid
32
Other existing Gateways The Virtual WindTunnel,
VWT 2.
Velocity vectors around a sphere in an
unstructured grid
PRELUDE can display results for both structured
and unstructured grids. It uses the public-domain
VTK file format.
33
Other existing Gateways Terrain
The Terrain Gateway facilitates simulation of
wind flow, air pollution and fire spread over
natural and human-built territories.
Thereafter users load terrain-description files,
wind profiles, pollution sources, choose grid
type, etc. Then PRELUDE write Q1, activates EARTH
and displays results.
Examples follow....
34
Other existing Gateways Terrain
35
Other existing Gateways Terrain
Here PRELUDE displays other aspects of the same
case. It can handle both structured and
unstructured grids.
36
Other existing Gateways Terrain
37
Other existing Gateways Terrain
38
Other Gateways under consideration
SHELLFLO considers only the shell of single-pass
shell-and-tube heat exchangers.
Many more configurations require study (see
right) and with boiling and condensation also.
For which would there be customers in Japan?
CHAM would welcome Japanese collaborators
in new-gateway creation.
39
User-Group 3 Users with new problems of their
own to solve

When no Gateway exists, and the problem to be
solved has novel elements, users need to find for
themselves what PHOENICS facilities to use.
Three examples will be referred to 1.
Data-centre cooling 2. Air-cooled condensers 3.
Populational CFD
40
User-Group 3 The Data-Centre problem
Its nature is simply Heating and Ventilating.
The difficulty lies in its magnitude.
Many computer cabinets and heat-extraction
units are crowded together.
.
But a parameterised Q1, linked to Excel, does the
job.
41
User-Group 3 The Data-Centre solution
PHOENICS provides graphical and numerical
results.
The easy-to-modify input-file structure allows
centre managers to optimise the location of
equipment units, and to detect dangerous
excessive-heating locations.
The data-input package was constructed by CHAM
UKs consultancy team for its own use but it can
be leased by others.
In the future, a Data-Centre Gateway may be
created...
42
User Group 3 The air-cooled-condenser problem
Air-cooled steam condensers involve three
phases air, steam and water, and three
geometrical dimensions.
There are size disparities between the total air
space and the thin gaps between tube fins.
The air occupies the whole space but steam and
water only a small fraction of it.
Unthinking reliance on the VR Editor would lead
to expensive and still inaccurate computations.
43
User Group 3 The air-cooled-condenser problem 2
The flow inside the tubes of the bundle may be
complex yet it greatly affects performance and
therefore must be simulated.
In cold weather, ice may form within the tubes. A
computer model should predict its effect.
The tube inclination will entail maximum
numerical-diffusion inaccuracy if the
easiest-to-use structured Cartesian grid (see
left) is used.
44
User Group 3 The air-cooled-condenser solution,1
Flexible PHOENICS permits unconventional
solutions, e.g. represent one volume twice, in
different parts of the grid.
This solves the numerical-diffusion problem.
It also allows one set of variables to serve for
two materials, viz air in one part and H2O in the
other.
Two extra additional grid parts would allow steam
and condensate to have distinct velocities and
temperatures.
The heat transfers between the air, steam and
water can be easily expressed by In-Form
statements in the Q1.
45
User Group 3 The air-cooled-condenser solution, 2
Both Space-Averaged and Detailed-Geometry CFD are
needed for the handbook data for
finned-tube-bundle heat transfer and pressure
drop are unreliable.
The PHOENICS PARSOL technique (see above) is very
suitable for the DGCFD part of this problem.

• PILIn-Form proved powerful enough
• to link object position and grid distribution to
  tube angle, length and diameter and
• to enable the thermodynamic properties of
  steam/air/water to be computed via statements in
  Q1.

46
User Group 3 The air-cooled-condenser
conclusions
The work, still in progress at CHAM UK, has
proved the multi-part grid to be practicable. It
is even better when inclined as shown here.
3D effects caused by the circular-fan shape are
small.
Those caused by the steam and condensate pipes
are larger.
It is too early to create a Gateway.
47
User Group 3 Populational CFD
Populational CFD treats multi-phase and
single-phase turbulent flows in a similar manner.
In my opinion it represents the new frontier of
CFD research.
PHOENICS is the computer code which is mainly
being used for its investigation
48
How Populational CFD differs from Conventional
CFD, 1
Both discretise space and time by use of grids of
cells, structured or unstructured.
Both solve algebraic mass-, momentum -
energy-conservation equations by iterative
numerical methods
Both take account of (1) sources, (2) diffusion,
(3) convection and (4) time-dependence.
49
How Populational CFD differs from Conventional
CFD, 2
Conventional CFD accounts for the four processes
just listed but Populational CFD accounts for
two more, namely
(5) Merging, by way of collision,
coupling-and-splitting or engulfment, which
influence turbulent combustion, and
(6) differential (i.e. selective) convection,
which influences buoyant and swirling flows.
50
One aspect of populational CFD Proportion-of-time
calculations
Calculated proportions of time that temperature
lies above or between given limits in flow behind
a square-sectioned obstacle in a non-uniform
temperature field.
A two-dimensional LES simulation was judged
sufficient for the present purpose. Typical
instantaneous temperature contours are on the
left.
On the right are contours of computed PTA_0.5,
i.e. proportion of time spent above a temperature
midway between highest and lowest.
51
Concluding remarks about populational CFD
For more on this research-frontier subject see
Mapping Turbulent Combustion at
http//www.cham.co.uk/lectures.php
52
PHOENICS-for-Teaching. PFT User-group 4
Professors teach students about fluid dynamics,
heat transfer, solid mechanics, etc., by lectures
and laboratory classes.
PHOENICS-for-Teaching is designed to help them.
PRELUDE and VR-Editor present too many choices
for this User Group 4. So PFTs
simple-as-possible GUI focusses on one topic at a
time, for example
53
PHOENICS-for-Teaching. PFT The Main Ideas

• What is to be taught can be classified as
• Subjects, e.g. Heat Transfer
• Sub-divisions, e.g. Convective heat transfer
• Further sub-divisions, e.g. extended-surfaces
• particular topics, e.g. finned tube In cross-flow

Professors, and text books, teach the
classification and classical methods of analysis.
PHOENICS-for-Teaching assists them at the
particular-topic level.

• PFT also handles what classical analysis can not,
  viz.
• practical geometries
• temperature-dependent properties
• complex interactions between processes.

54
PHOENICS-for-Teaching. PFT The simple user
interface
The interface describes the case then allows
data input.
55
PHOENICS-for-Teaching. PFT The user interface
56
PHOENICS-for-Teaching. PFT The user interface
concluded
And finally
Thereafter the student presses run the
simulation then views the results graphically,
by means of macro-driven PHOTON or VIEWER or in
alphanumeric form.
What could be simpler?
57
PHOENICS-for-Teaching. PFT A future
collaborative development
The PHOENICS-for-Teaching Project has just begun.
Therefore only a few topics have been dealt with
so far.
However, the structure exists and the way ahead
is clear so progress may be rapid.
CHAM will supply PHOENICS-for-teaching free-of-
charge to collaborators who agree to make their
files available to all teachers and students
world-wide.
58
User Group 5. Four developments which may
interest experienced users of PHOENICS
1. A PARSOL and Parallel problem solved.
Parallel PHOENICS works by splitting a large
domain into many slightly-overlapping sub-domains
and assigning a processor to each.
When PARSOL is used, neighbouring processors
sometimes disagree about cell-cutting in the
over-lap region, causing execution failure.
The solution has been Let them disagree but
let the second processor accept the first
processors decision.
Thats easily said but hard to implement in
coding. Nevertheless, the problem has been
finally solved.
59
User Group 5. Four developments which may
interest experienced users of PHOENICS
2. A welcome discovery about GCV
For wind-farm simulation, users of Windsim, the
user-interface package for which PHOENICS is the
CFD engine, have often encountered divergence.
Windsim uses a BFC grid to describe the terrain
and had used the settings MIGALT GCVF.
CHAM then advised set MIGALF GCVT whereupon
convergence was obtained unfailingly.
The conclusion? GCV (General Colocated
Velocities) should always be selected for BFC
problems.
An objection is the GCV solver has not been
parallelised. This omission is being remedied.
60
User Group 5. Four developments which may
interest experienced users of PHOENICS
3. Automatic convergence
Wind-farm (and general terrain) problems are
comparatively simple (small height/width ratio
fixed-pressure at sky) so procuring convergence
is not hard.
For general flows, CHAMs CONWIZ automatic-
convergence promoter is usually effective but it
is being improved in a current study (not yet
complete).

• Findings so far are
• Never use false-time-step relaxation
• Residuals values are unreliable tests of
  convergence.
• Small LITER for velocities (even PBP! are
  advisable.
• Optimal LITER(P1) is of ten less than 100.
• Further systematic studies are needed.
• Guaranteeing convergence is the duty of CHAM.

61
User Group 5. Four developments which may
interest experienced users of PHOENICS
4. Improved graphical-display features
Such improvements are too numerous to list here.
Therefore I conclude with one so small that it
may escape attention controlling GXMONI line
thickness.
For many years GXMONI lines were one-pixel thick.
Now, at last, user can change the thickness via
cham.ini.
The image on the right shows three monitor plots,
with LineWidth 10, 4 and 2.
62
New Trends in PHOENICS Development a final
summary
PHOENICS appears now differently to each user
group.
5.Traditional users see it as changing only in
detail and in richness of options
(structured/unstructured, etc).
4. Student teacher users see an easy-to-handle
mini-PHOENICS, which does what they need and no
more.
3. Innovative users at the frontier of CFD and
its applications see PHOENICS as a box of power
tools but they may need help in using them.
2. Gateway creators, learning to write
parameterised Q1 files, automated by PRELUDE,
assist their front-line colleagues.
1. Then the latter can be productive and
trouble-free.