You are on page 1of 4

Synopsis of the Transient solvers in ESATAN.

1. SLFWBK

This routine performs transient thermal analysis by implicit forward-


backward differencing—the Crank-Nicolson method. EXPLAIN THE CRANK
NICOLSON METHOD
The user is required to specify the following control constants: the time
step length
(DTIMEI); the problem start time (TIMEO) if this is not 0.0; the problem end
time
(TIMEND); the output interval (OUTINT); the maximum iteration count
(NLOOP)
for each time step; the relaxation criterion (RELXCA) for convergence on a
time
step.
The time step length is taken as the user supplied initial time step length,
DTIMEI, unless the control constant DTPMAX has been set. DTPMAX
specifies the maximum temperature change allowable over a time step.
The program increases the time step to bring the maximum temperature
change in line with the user supplied maximum. If DTPMAX is exceeded
then the time step is reduced and the temperatures recalculated. If
DTPMAX has been set to regulate the time step length then it is advisable
that DTMAX and DTMIN are set. DTMAX and DTMIN specify the maximum
and minimum time step lengths allowable. Where DTPMAX would lead to a
smaller time step, DTMIN is used and the solution continues but an error
code is set. The change in temperature at each node may be damped via
the control constant DAMPT.
The user can specify the type of units used by setting the control
constants STEFAN and TABS.
$VARIABLES1 is executed at the start of each time step and $VARIABLES2
at the end. $OUTPUTS is executed at the beginning of the transient, every
OUTINT units of time during the transient and on completion of the
transient.

Restrictions:
This routine can only be used with pure thermal models. Control constants
DTIMEI, TIMEND, NLOOP and RELXCA must be specified. It requires 3*NNT
locations of dynamic core (NNT = total number of nodes in system).
SLFWBK
may be called from the $EXECUTION block only.
2. SLCRNC

SLCRNC performs the transient by calculating the rate of change of


temperature of
each node at the beginning and end of the time step, and using the
average to
predict the end of time step temperatures (the change in temperature of a
node
being the average rate of change multiplied by the time step length). The
variation
in the rate of change of temperature over time is an important indicator of
acceptable time step length, and hence DTROCA specifies the limit on the
maximum change in the rate of change of temperature over a time step
multiplied
by the time step length. In most circumstances a value of between 1.0 and
10.0 for
DTROCA is appropriate.
3. SLGEAR and SLGRDJ

These routines perform transient thermal analysis by backwards


differencing using
the Gear formulation. The time step length is dynamically optimised by the
solver
after using an initial value of CSGMIN.
The user is required to specify the following control constants; the problem
start
time (TIMEO) if this is not equal to 0.0; the problem end time (TIMEND);
the
output interval (OUTINT); the maximum iteration count (NLOOP) for each
time
step and the relaxation criterion (RELXCA) used to check convergence and
to
choose time step length and solution order. The control constants DTMIN
and
DTMAX may also be set to specify the minimum and maximum time steps
allowed.
The user can specify the type of units used by setting the control
constants
STEFAN and TABS.
The Gear formulation consists of a prediction followed by an iterated
corrector
stage for each time step. This corrector stage involves solving a matrix
equation to
yield a vector of temperature increments to be added to the prediction.
SLGRDJ
differs from SLGEAR in that the off-diagonal terms of the matrix in this
equation
are ignored. In effect, then, SLGEAR solves the temperature increment
equation by
matrix inversion, and SLGRDJ by successive point iteration.
$VARIABLES1 is executed at the start of each iteration within a time step
and
$VARIABLES2 on completion of each time step. $OUTPUTS is called at the
start
of the transient, every OUTINT units of time during the transient and at the
end of
the transient.

4. SLMODE

This routine performs transient analysis using the modal analysis method.
Givens' method followed by the QR algorithm is used to evaluate the Eigen
values
and the Eigen vectors. The user is required to specify the following control
constants: the time step length (DTIMEI); the problem start time (TIMEO) if
this is
not 0.0; the problem time end (TIMEND) and the output interval (OUTINT).
The
user must specify a maximum iteration count (NLOOP) and relaxation
criterion
(RELXCA) for solution of zero capacitance nodes.
If convergence to within RELXCA is not achieved inside NLOOP iterations
for a
time step then the solution continues but an error code is set. The control
constants
DTMIN, DTMAX and DTPMAX may be set in order to regulate the time step
used
by the program. DTMIN specifies the minimum time step length allowed,
DTMAX
the maximum time step length allowed and DTPMAX the maximum
temperature
change allowed over a time step. Where these criteria clash, e.g. a time
step smaller
than DTMIN gives a temperature change greater than DTPMAX or DTMAX
set
less than DTMIN, DTMIN prevails and the solution continues. If DTPMAX is
exceeded because of this an error code is set.
The user can specify the type of units used by setting the control
constants
STEFAN and TABS.

You might also like