A
Simple Transient
Goals
- Introduce the SNAP Model
Editor.
- Show you how to generate
plots of results.
- Illustrate a limitation
of numerical simulations.
Preliminaries
Repeat the steps given
in Exercise 1
to start the calculation
server. If you are working a Windows rather than Linux computer
operating system, you will also
need to
start an X Server, to support the plotting program (AcGrace). In the ITS
computer laboratories from the Programs item of the Start menu, look
for Utilities, then cygwin, then the item to start Xwindows. If
you are on another computer with Cygwin installed look for an icon or
Program menu item labeled “Cygwin”. Double-click
the icon or select “Cygwin Bash Sell” from the “Cygwin” program
submenu. In the
window that opens, type “startxwin.sh &” or the long form
"/usr/X11R6/bin/startxwin.sh &", then wait for it to open an X
command window. As another option, you can use the Exceed
Xserver if it is installed. Look
for an icon or item
in the
Programs menu for “Exceed”, and start the Exceed X server by clicking
the
“Exceed” item in the Exceed submenu.
Be sure that you’ve
created an
“Exercise-2” subdirectory to store results from this work.
Problem Definition
Flow is
through a 10 meter section of 0.25 m diameter circular pipe.
Initially,
water
flows into the test section at 1 m/s with a temperature of 300K. From 1 to 2 seconds into the transient, the inlet
temperature
increases linearly from 300K to 330K. The pressure at the exit is 2.0x105
Pa.
For your model, divide the test section
into 10 equal volumes. Run a transient simulation for 10 seconds with a
time
step size of 0.1s, and graph the temperature in the 6th
volume
(x=5.5m) as a function of time. How does the result compare to the
exact answer
for flow without turbulent diffusion?
Procedure
As
with any flow
simulation, most of your work will be involved in setting up the
initial and
boundary conditions for the problem. You must first learn to use the
Model
Editor provided with the Symbolic Nuclear Analysis Package (SNAP), to
provide
input to TRACE. You will build input models through a combination of
menu
selections and keyboard entries. When
you complete a keyboard
entry in a box or table, be sure to hit
the “Enter” (Return) key, or click in another box, before you click an
“OK”
button to close the dialog window. If you don’t your last entry may not
register with SNAP.
Begin
by
selecting “SNAP” from the “Programs” menu. You will see the options
below.
Select “Model Editor”.
The Model Editor window will
appear on the screen, along with a Message window that allows you to
select your starting point for the model. We will be creating a
new model.
After clicking "OK", a small
window appears
in which
you elect to generate a TRACE input.
Clicking the “OK”
button opens three panes within the main SNAP window. Upper left
is a list of all major input model categories called the "Component
Navigator". Bottom left is space for detailed component
descriptions. The larger area to the right is called the "Default
View" where you will graphically build your system model. Later,
when you check or export your input model, a fourth pane will appear at
the bottom of the window containing messages . If you prefer a
format with separate windows for each of these panes, you can request
that in the Preferences item of the Edit menu.
Look at the tool bar
containing binoculars and padlock icons. You should see a button
with an arrowhead pointing down. If
you hold
your mouse cursor over the button without clicking, a label appears
reading
“Select component to insert.” Click this button, and a menu appears
from which
you should select “Hydraulic Components.” This generates a submenu on
which you
should click “Breaks” (pressure boundary condition).
The menus disappear,
and the first
button in the second tool bar appears as depressed. This is your
shortcut
button for getting “Break” components, and can be used directly to
bypass the
steps to the “Hydraulic Components” menu. Your cursor should now appear
as a
cross in the workspace of the Default View. Select a position in the
workspace
and click the mouse once. A break icon will appear. At the same time,
an entry for the new break should appear in your Component Navigator
window and a Properties Dialog should appear associated with that
window. We will return to the properties shortly.
Return to the “Select
component to insert” button twice more to add one “Fill” (flow boundary
condition), and one “Pipe” component. In both cases note the button
that
depresses in the second tool bar to make your life easier later. After
your
selection of a pipe, its representation will appear along with a window
asking
for the number of cells in the pipe. One strategy here is to
figure out in advance how you are going to divide space into volumes
for the calculation. In that case I would enter "10".
However, that is normally the hard route. I first think
about the geometry of the flow path. In this case I'm modeling a
10 meter long section with uniform flow area. I enter "1" because
the basic geometry doesn't vary over the length of the component.
If I was setting up the input for the pipe in Exercise
1, I would have entered "2" because it has two uniform
regions. Later we will split the pipe into more volumes to get
better results from the flow equation solution.
Be very careful when
in component
selection mode. More than one click on the workspace will produce more
than one
copy of the currently selected component. You may
want to take a minute to
clean up
the picture. Click the general selection button 
,
and drag the Fill
to the left of the Pipe and the Break
to
its right. Right click on either the fill or break to
obtain a menu option to adjust the orientation to something like I've
got above.
Next you must
establish flow
connections between your components. Select the “Connect two
components” button 
. Click the modified cursor first
on the marker at the right end of the pipe, and then click it again on
diamond at the bottom edge of the break. A blue connecting line should
appear.
Repeat the process to connect the small circle on the tip of the Fill
to the marker at the left edge of the pipe. If at some point a
small dialog box opens asking for a connection type select the Hydro
Connection not the Leak Path.
You’ve finished the
basic layout
of the system. This is a good time to save your
work. Select “Save” from the “File” menu on the Model Editor’s
main menu bar,
and navigate through the “Save in” dialog to drop the SNAP data file
into your
Exercise-2 folder.
The next job is to set
the
geometry and initial conditions for all of the components. Start by
moving your mouse cursor into the
Default View (screen shot above) then right click the picture of the
pipe and select "Properties. You can access the same dialog by double clicking the pipe's entry in the Component Navigator.
Start in the Properties View by giving the
pipe a short
name, a description, and some relevant comments. To make entries
in the Description and Comments fields you need to click the
button to the right with the red E
and diagonal arrow. 
Now move on to the
Component Geometry and again click the data expansion button .
You will get a separate data entry window with tabs labeled "Cells" and
"Edges". First select Edges, ad check to see that the hydraulic
diameter has automatically been set to 0.25. If you need to
make a change, click in the appropriate box, or drag your mouse across
a series of boxes that will receive the new value, and type. When
you
hit the Enter key that value should appear for both hydraulic
diameters, and if the flow area was set to 'Unknown", then values of
the flow area should be computed based on a
circular cross-section. If this were a rectangular duct or some
other odd shape, you could overwrite the area values.
For this problem we
are using a horizontal pipe so don't touch the Orientation for
now. You can play with that later if you have time. In this
problem the only noticeable difference orientation will make is in the
calculated cell pressures during the transient. Move on to
the Cells tab. Take a look at volume properties. They
should have been automatically calculated based on a straight
cylindrical pipe with the hydraulic diameter that you specified when
the pipe was first created.
Now go back to the general pipe
properties
dialog, and expand the Component Initial Conditions. Get values for the
state variables from the Problem Description above. The value
that you enter for the vapor temperature won't matter, because there is
no vapor in the system
Click the Close
button and move on to the break properties dialog.
Standard guidelines for breaks and fills are to set the cell length and
volume equal to the adjacent active fluid volume. I know that I'm
going to split the pipe into one meter long cells, so I set the break
length to 1 meter, and multiply that by the pipe cross-sectional area
to get the break volume. Set the Pressure to
2.0e5
Pa, and Mixture Temperature to 300.0 K. Leave everything else
unchanged. This
is a very simple constant pressure boundary condition.
Now double click the
Fill representation in the
Default View
workspace. Do the Identity category as before and move on to the
Options. The
Fill Type should be set to “Generalized State
Table”.
The Cell Length and Cell Volume should match those in the first cell of
the
pipe. Set remaining properties to values appropriate for this
problem.
To finish the job
you've got to setup the Fill Table. We want a table that keeps
liquid temperature constant
from 0.0 to 1.0 seconds, increases it linearly to 330.0 between 1.0 and
2.0
seconds, and holds it constant for times greater than 2.0 s. Expand the
Fill Table entry and select the independent variable to be the problem
time. This is your only choice, but more will be available when
we create problems with signal variables and control blocks. Next
click the Add Row button. The independent variable is
time, and should start at 0.0. The remaining state variables
should be set according to the problem definition.
Click the Add Row
button again.
Set the Independent Variable to 1.0, and verify that all other entries
match
the first row.
Click the Add Row
button again.
Set the Independent Variable to 2.0, the Liquid Temperature to 330.0,
and
verify that all other entries match the first row.
Click the Add Row
button again.
Set the Independent Variable to 10000.0 (big number), and verify that
all other
entries match the third row.
Now we have the basic
initial
conditions, all that is left is to define some basic information
describing the
overall calculation. In the “Component Navigator” window, right click
line labeled "Model Options", then click “Model Properties” in the
drop-down
menu. In the properties box, enter useful information for the Model
Name, Title Cards, and Model Description. The Title cards are
actually read by TRACE and printed to its output file. Set
"Transient Calculation" to "Yes". Accept defaults on
other direct entries.
Move on to the
“Timestep Data” and expand it. You’ve been asked to run the
calculation for 10 seconds, with time
step sizes of 0.1 seconds. Set the “End Time” (tend) to 10.0, and
the
“Maximum Size” (dtmax)
to 0.1. You can leave the “Minimum Size” (dtmin) at 1.0e-6 if you like.
When the code
has trouble getting a solution, it can automatically reduce the time
step size.
If it tries to cut the time step below “Minimum Size”, the code quits.
My
experience is that anytime TRACE tries for a step below 1.0e-5 seconds,
there
is something wrong with TRACE that should be reported to code
developers. If
steps stay below 1.e-4 for a prolonged period the problem should also
be
studied by code developers.
Set the ”Long Edit Interval”
(edint) to 1.0 seconds to give a few detailed
edits during the
transient. Set the "Graphics Interval" (gfint) to 0.1 to give values in
the
graphics file (trcxtv)
every tenth of a second. We will worry about other settings on this
dialog
later, for this exercise accept the defaults.
The last category that
we need to change in the
Model Options dialog is the Namelist data. This is often a mystery to
beginning user’s even with the
description in the TRACE Users’ Guide. Normally, you want the default
settings.
In the dialog’s table find the name “DTSTRT”. Without intervention, the
code
will start a calculation at the “Minimum Step”, and gradually build up
to the
“Maximum Step”. We’ve got a gentle enough calculation that it’s worth
starting
with a time step of 0.1 seconds. Check the active box before “DTSTRT”,
set the
Value box to 0.1, hit the Enter (Return) key, and click the OK button.
Save
your project from the Model Editor file menu.
Now we in
principle have a complete model
description. However, using one volume to represent the full pipe
won't give much resolution of the thermal wave propagation. Right
click on the pipe's picture in the Default View or it's entry in the
Component Navigator, select "Special" from the drop-down menu, and then
select "Renodalize." Click on the picture of the pipe so that it
changes color, and the two Split buttons become active.
Click the "Split
Uniform" button and tell it to
split into 10 cells. Close the window by clicking OK, then take a
look at the pipe property dialog to see the change in cell geometry.
At this point, run the
transient from the “Submit Job” item on the Model Editor “Tools”
menu. Turn on the Job Status tool to check the progress of the
calculation and view data.
To get a graphical
view of temperature, select
“Plot” from
the Tools menu Job Status application. This will only work if you
remembered to start your X
server at the beginning of the session (not a problem for Linux
users).
To read the
TRACE
graphics file, go to the AcGrace “File” menu, select “Read”, and
on the
resulting
submenu select “TRAC data”.
A dialog now appears
that lets you locate your
graphics
file.
. 
In the menu list above labeled "Directories" select the directory
containing the results of your calculation (check the list of Job names
in the Job
Status window if you aren't certain). Next click in
the Filter button. At this point you
should be
able to find and double-click “trcxtv” in the “Files" list. Once the
“Selection”
box shows the full path to your graphics file, click the “OK” button,
and the
following dialog appears.
Select “pipe-021” in the first box with a single
click,
tln-021006 in the second box, and click the “Accept” button. You should
see a
plot of temperature versus time in the 6th cell. How does
that
compare to the correct answer? How would it affect conservatism of a
thermal
stress calculation?
Another Way to Use Tables
Tables such as you entered for the fill are available in many other
components for various purposes. However, they are not
always the best approach. You probably noticed that most of the
information in the fill table above was for state variables that
remained constant. In other components a more serious limitation
exists when restarting a calculation. For most complex
simulations you will find yourself developing a full input model only
to establish an initial steady state. After that you pick up most
of the system information from a restart file dumped by TRACE ( trcdmp
or PREFIX.dmp), only changes necessary for a specific transient
scenario are introduced in the ASCII restart input file created by the
SNAP Model Editor (or by hand). If you want to run a set of
transients from the same steady state with different time histories for
a valve area, it is more trouble than necessary to extract the entire
state of the valve for inclusion in all transient inputs. There
is always a way to move the table information to a "function" control
block,
and only replace the control block at restart.
For most component tables this is just a question of picking a
component option requiring a table, giving zero as the number of table
entries, and supplying the identifier of the function control block for
the component table's independent variable ID number.
However, within the Model Editor you will
first need to create a function control block to drive the fill before
dealing with the fill properties.
In the Component Navigator window right click the "Control Systems"
item and select "Create View." The Model Editor always gives you
a signal variable for transient time, which you should see in this
view. You may want to use the View menu of the Control Systems
View window to zoom things a little. You
should also see a button on this window's General Toolbar with an
arrowhead pointing down. If
you hold
your mouse cursor over the button without clicking, a label appears
reading
“Select component to insert.” Click this button, and a menu appears
from which
you should select “Control Systems.” This generates a series of two
submenus where you should select "Control Blocks" and "Arithmetic
Blocks." Position the cursor to the right of the Time
signal variable, and click to initially display the control bloc, use
the connection tool
to
connect the two.
You will need to edit the control block's Properties dialog,
changing the type to "(102) Function" and adding documentation.
Finally expand the "Function Table" item, click the "First" button to
enter the first and only independent variable, and enter the four times
used in the previous fill table.
Click OK in the Edit window, then add each temperature in the column
labeled "Y".
Now you are ready to change the fill to
work through a different fill type. Open the fill's "Properties
View" and select a Fill Type of 10 , then click the "+" next to
the "State Controllers" line to expand its view.
In this problem only the liquid temperature changes, so just click the
selection button 
in the
row labeled "Liquid Temperature Controller", and select the control
block that you just created.
Finish by setting the flow rate for the fill. A type 10 wants a
mass flow rate rather than a velocity. Look at the output from
your first model to get the liquid density in the fill and the flow
area. Multiply the density, flow area and liquid velocity
(1 m/s) together (48.94 kg/s) and enter the result in the fill property
view. At the moment the type 10 fill doesn't accept the
mass flow, so you need to go to the Model Editor window and from
the File menu, select "Export", then "TRACE", then "ASCII Full", and
finally provide a name for the input file. Open that input file
with jEdit, locate the label "flowin", and change the number
below it from 0.0 to 48.94 . Run the problem from jEdit,
and compare your results to the first version of this model. They
should match.
This is all we are going to do with control blocks for a while.
You will get a chance to use this powerful feature in more depth later.
Created and Maintained by John
Mahaffy : jhm@psu.edu
©2004