CHT12 - Phoenics

lilactruckInternet and Web Development

Dec 4, 2013 (3 years and 9 months ago)

61 views

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


Computational Heat
Transfer

in Engineering Education


Keynote lecture at

International Symposium

CHT
-

2012


by

Brian Spalding

of CHAM Ltd

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


Some preliminary ideas



Which is easier to
understand
?

A

finite
volume
?

Or an
infinitesimal

one
?

Then why do we teach
students about
differential
equations

before
finite
-
volume
ones
?

We should help it to do
so.

Because finite
-
volume thinking
has not yet trickled down to
pre
-
university level.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Upwind
-
differencing eluded
mathematicians for many years.

The tank
-
and
-
tube concept
embodies finite
-
volume thinking

fluid on the
upstream

side of
the ‘tubes’ flowed
in

to the
‘tank’;
fluid
from

within the tank

flowed
out.

Not for the first time, physical intuition turned out to
be at least as productive as mathematical training.

To ‘tank
-
and
-
tubists’ it was
obvious
:

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


Intuition suggested that
grids made of cells like
this might reduce
numerical diffusion.

A later tank
-
and
-
tube idea:
the X
-
cell

It was right

Square
-
cell grids smear
badly in diagonal flow

Doubling the number of
square
cells has much less
effect

In the same flow
X
-
cells do not
smear at all.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Solving

simultaneous
differential equations was hard.

A 1954 example of
finite
-
volume success
:
1
-
D flame propagation

even in graphical
Schmidt
-
method form!

Temperature profiles on
the right show:

(a) response to the
initial discontinuity;

(b) Temperature
-
dependent sources
then promote flame
propagation.

Dusinberre’s 1949 book used
finite volumes
for
heat
-
conduction
; so why not use them for
combustion

too?

Finite
-
volume
thinking had
solved the
problem
. .

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Heat transfer is taught by way of differential equations
because (a few)
analytical solutions exist
;

A summary of the argument to
be presented

but
only

for seldom
-
realistic conditions (
uniform
heat
-
transfer coefficient, temperature
-
in
dependent properties).

If

the solutions are used for design,
large ‘safety
factors’

must be applied.

Therefore serious designers
would

use
finite
-
volume
-
based
computer simulation instead, but only
if

Therefore CHT specialists have a duty: to

promote the
finite
-
volume

formulation
throughout
education

This has
world
-
wide

(and bad) economic significance.


they recognised that industry
-
standard design software
is
still

based on the
unrealistic

presumptions.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Today’s teachers
are

‘computer
-
savvy’,
using lecture
-
room
computers, Power
-
Point, GOOGLE
searches,
etc

Education through simulation
in the lecture
-
room

But few perform live
simulations
:
they lack tools,
knowledge

and
confidence.

My proposal is that, for
heat
-
transfer

lecturers,
CHT

can
and should
provide all three
. I shall explain how.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


Example:

a file for
flow
&

heat transfer
in

tubes
;

What the CHT community could
provide:
down
-
loadable

‘Simulation
-
Scenario
’ files

d

The main window contains the start of an
html file
,
which the lecturer is free to edit .

Above are buttons enabling him/her to do
live
simulations
of flow in tubes.

which a
‘SimScene’
-
viewer
package reads;
and then shows this


interactions
with which
constitute the lecture.

But the
input data
must first be inspected.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Clicking on the
appropriate bar
and then on the
seventh left
-
hand
box causes this
menu to appear.

About the TubeFlow package:
data input
via

menu

The
flow formulations

which
can be chosen are shown here.

The lecturer may choose to explain their meanings,

perhaps after first studying the html document.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Heat
-
transfer text
-
book

formulae
connect Nusselt,
Reynolds and Prandtl numbers, each containing
thermo
-
physical property values, treated as
constants
.

Handling temperature
-
dependent properties

In reality, properties
vary with temperature.

A SimScene
-
using
lecturer can explore
these effects
in the
class
-
room;

and
more easily
than
in the laboratory; both
real and fictitious fluids
can be investigated.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Just click on the running man icon; then in a few
seconds the results are available to be displayed.

How the in
-
classroom
simulations are performed

Nor need the lecturer know how to operate the graphical
display package; for, when the run ends, he will see this:

Clicking on the icon will
activate a macro which creates
images automatically, such as:

contours of temperature

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Some merits of the
down
-
loadable SimScene system

Graphical displays
make more impact on students’
minds than algebraic derivations.

Moreover students can make

explorations for
themselves
in ‘SimScene homework’ sessions.

and

will have learnt that CHT/CFD
has limitations
too:
-

Later, as
professional engineers
, they will be readier
to use
finite
-
volume
-
based

simulation for design;

viz.
grid
-
fineness

effects;

computer
-
size

needs;

and
human error
.

turbulence
-
model uncertainties;

The lecturer needs only minimal computer skills; and
he/she can deliver ‘
as is
’ or with own

embellishments
.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




The task: Use TubeFlow’s
multi
-
run

capability
to compute fluid flow and heat transfer
for water,
at 80 degC, in
fully
-
developed flow
, for
various Reynolds
numbers; and explain the results


Example of a homework
assignment

The student might
obtain this


Not bad. But how
explain the
drooping of the
Nusselt No curve

(bottom
-
right)


Unravelling puzzles promotes understanding.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


Lecturers should
know enough about CHT
to explain
its sources of
uncertainty
:
too
-
coarse

grids,
turbulence

models,
multi
-
phase

effects.

Exploring the influence of
uncertain inputs

Then they can enlarge their
students’ knowledge

(and
their own) by saying: ‘Run each turbulence model; then
compare results’.


TubeFlow makes this easy. Here is its multi
-
run screen:


It will launch 30 runs: 5 models for each of 6 velocities.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Further advantages: some things
need no longer be taught

Text
-

and hand
-
books are cluttered with formula which
purport (implausibly and impractically) to be useful in
design. For example:

Likewise, figures like this,


with impossibly low Nusselt
Nos.

They represent someone’s
long
-
ago hopeful guess; and
they are copied from book to
book without criticism.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


.






More fiction to be binned

Finned
-
tube bundle Nusselt and Euler Number
formulae according to Rohsenow and Hartnett:

They are neither reliable nor credible because ...

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


(
1
)

The

number

of

dimensionless

parameters

needed

for

finned
-
tube

bundles

should

be

at

least

12

.


The reasons for binning

(
2
)

The

army

of

experimentalists

needed

systematically

to

explore

this

12
-
dimensional

space

has

surely

never

been

mobilised
.


Nor

will

it

ever

be
.

(
3
)

Even

if

it

had

been,

it

is

highly

improbable

that

its

findings

would

have

fitted

the

always
-
preferred

form
:

Nu=a*
Re
b
*
Pr
c
*D
e
*F
g
*H
i
J
k
*L
m
etcetera


wherein

a, b, c, e, g,
i
, k
and

m
are constants, and
D, F,
H, J
and

L

etcetera

are dimensionless parameters.


Only
a SimScene package
devoted to
finned
-
tube
-
bundle geometries can work out
the interacting influences of all parameters.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Step 1. Decide
what

parameters
define the scenario,
e.g.
shapes, sizes, materials, thermal conditions
.

How to create a SimScene
package; what’s involved?

Step 2. Decide

what default

values (or lists) shall
appear in the SimScene
-
viewer’s menu boxes.

Step 3. Decide
what CFD engine
will perform the flow
-
simulating calculations.


Step 4. Express the above decisions in the CFD
engine’s
Data
-
Input language
.

Comments: (a) Steps 1 and 2 are the
creative

steps

(b)
Re
Step 3,
any

general
-
purpose code will serve.

(c) Step 4 requires knowledge of the engine’s language;
but it is
mechanical
in essence.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




1. Knowing only the PHOENICS Input Language (
PIL
), I
used it; but repeat: SimScenes can use
any

CFD engine.

How to create a SimScene
package; some details

2. There does exist a
PIL editor
, with macros, widgets
and other aids. They
may exist

for other engines too.

3. It is Steps 1 and 2 that require
agreement on format
.
Commercial competition should not hinder its making.

4. The current SimScene format can of course be
improved; but refinement
is cheaper than
replacement.

5. For those who make the same choice of engine, I can
provide the Editor,
and how
-
to
-
use instructions.

6. My aim is to bring into existence a ‘
critical mass
’ of
SimScenes in a short time. Can that be made

possible
?

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




“I take it those things are to be held
possible

which may
be done by
some person
though not by
every one
;

Regarding
possibility,
Sir Francis Bacon

wrote:

“and which may be done by
many
, though not by any
one;
.

“and which may be done in the
succession of ages
,
though not within the
hour
-
glass

of any one man’s life;

“and which may be done by
public designation
, though
not by
private endeavour”.

What
public
? Could that be the
CHT community?
Perhaps in

co
-
operation
with like
-
minded others?

For, if
‘many’
participate, a plethora of SimScenes might
exist sooner than
‘succession of ages
’ suggests.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




Its top page looks
like this:

Another heat
-
transfer
-
related
SimScene: HeatEx

If

a
ll SimScenes:
have similar forms,
novelty of

content

stands out better.

Author, date and
institution would be
useful additions.

.

If a CHT
-
SimScene
-
Creators’ Club came into existence,
a first task would be to recommend an all
-
fitting format.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




HeatEx is designed to teach students about shell
-
and
-
tube heat exchangers, like this one:

The purpose of HeatEx

Influences of tube and baffle number and positioning are
among those to be simulated.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




As well as the text
-
book
-
standard options,:

parallel
-
,
counter
-

and
cross
-
flow, it has
oblique
,
two
-
baffle
,
leaky
baffle and

four
-
baffle options.

Available flow configurations

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




It has a menu structure similar to that of TubeFlow.

Data
-
input facilities

Here the flow configuration is being selected.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




For economical programming and display, it uses
two

grid segments to cover the same space: one for the
shell
-

and one for the
tube
-
side

fluid.

Its multiple
-
grid feature

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


Simulations run in the classroom by the
lecturer

or as
home work by the

students
reveal:

What students can learn from
the HeatEx SimScene

1. That finite
-
volume
-
based simulations
fit the text
-
book
formulae
closely enough
, if the grid is
sufficiently fine.

2. That contours of
temperature in cross
-
flow exchangers may
look like this


3.That baffles bring
close
-
to
-
counterflow

effectiveness

but raise
the
pressure drop.

4.That phase
-
change effects can be taken account of.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer




A few such packages exist; and I hope to create more.

Last words about simulation
-
scenario packages

But
hundreds
are needed, for the teaching of all
relevant fluid
-
flow, heat/mass transfer topics; and my
personal ‘
hourglass
’ will certainly not suffice.

I hope therefore to have
conveyed the vision
clearly
enough for some of
you

to share it; and to desire to turn
it into reality.

Finally, I disclose that there is a
sociological
aspect to
my ambition: properly considered, the

‘trickle down
’ of
finite
-
volume thinking into secondary schools can
widen
the entrance doorway

of the engineering profession.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


The

only
mathematics which is
essential
for understanding CHT is
that of the
storekeeper;

The last slides

‘Infinitesimal’ is
like
‘truly fine
enough’
.

So it no longer makes sense to
bar from studying
engineering

university entrants for whom calculus is
too high
a hurdle
.

Beyond our reach!

and he has

computers
to keep his
books for him.

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


A final speculation

We laugh because
medieval scholars
debated how many
angels could stand on
the head of a pin.

One day we may scorn

mathematicians for their
obsession with the
infinitesimal.

But
finite
-
volume
-
based
simulation will survive!

Bath

July 1
-
6

2012

Advances in

Computational Heat Transfer


and to
all of you
for your attention.

Thank you

To
Alexey

Ginevsky

for conceiving and creating the
SimScene
-
viewer package and the PIL
-
editor;

to
Elena
Pankova

for her work on
TubeFlow

and
HeatEx
;

The End