Modifications for V 2.0
May 1995
MISES V 2.0 is a fairly major rework from V 1.4.
Specific major changes are:
A) The inlet/outlet condition specification has been extended
and conceptually simplified. The inlet/outlet conditions can
now be specified anywhere in the domain, not just at the grid
inlet/outlet planes. The outlet condition is also on a true
mixed-out state, rather than some ad-hoc mass-averaged quantity.
The mixed out losses are also computed from this mixed-out state.
The ises.xxx input file has a new format. The small program
isconv.f will overwrite an "old" ises.xxx file with the equivalent
"new" ises.xxx file.
Usage:
% make isconv
% isconv xxx
B) A modified version of the Abu-Ghannam--Shaw bypass transition model
has been implemented. The standard AGS model has been found to be
ill-posed (see writeup in ./doc/ags.ps ). For high to moderate
turbulence levels in favorable pressure gradients, this will typically
cause transition before the standard e^n envelope method. In fact,
bypass transition can occur before e^n (TS wave) instability sets in.
The implementation was an ad-hoc addition to the existing e^n formulation.
When the local Rtheta approaches the AGS critical Rtheta, a very rapid
fictious growth rate is added to the usual e^n amplication rate so that
transition is achieved within a small number of BL thicknesses.
C) The internal boundary layer routines have been generally "cleaned up",
and now reside in their own source directory ./blsrc . There is now
one general BL access routine MRCHBL (in blsrc/mrchbl.f), used by MISES
in various ways. This routine can also be run as a conventional BL code
independent of MISES. A full call list description is in the routine's
comment header.
Modifications for V2.1
A) The block matrix solver has been replaced with a "vector-skyline"
solver which accepts a fairly general matrix form. This now allows
general offset grids over the inlet and outlet.
B) The old arcane grid generator has been replaced with a much
more friendly version adapted from the multielement MSES code.
A 2-D panel solution is used to generate the initial streamlines
and to properly block-off the offset grids and the inlet and/or outlet.
C) A new ISMOM=4 option has been implemented, which makes the flow
isentropic everywhere except where dissipation is active. Losses are
now generated only where they should be generated -- i.e. at shocks.
Subsonic regions are guaranteed to be isentropic.
D) Streamtube-thickness variation modes have been implemented.
It is possible to have ISES implicitly determine the outlet
streamtube contraction to match a specified outlet pressure,
for example.
E) The idat.xxx file now contains only the minimum amount of
information needed to define the case flowfield. Previously,
the entire contents of STATE.INC were written out up to the array
dimensions, which made idat.xxx unnecessarily large for smaller
cases. Secondary variables like Q(I,J) and PI(I,J) are also
no longer written out. Now there is little drawback to making
the primary array dimensions in STATE.INC comfortably large.
F) The treatment of Mixed- and Modal-Inverse parameters in EDP
has been cleaned up. The manual has a more extensive writeup
on these two approaches.
June 18, 1995
Most everything seems to be in order. All codes compile on
DEC, IRIS, HP, RS machines with no complaints other than
a few nuisance warnings. The DEC Alpha OSF compiler failed
on a few of the source files with -O2 and above. Beware
of over-zealous optimizers!
The following cases have just been checked out:
uti
t7
dfvlr
The arltandem is experiencing erratic behavior on the splitter
wake at the outflow boundary. I'm not totally happy with the
exit BCs there, but any fix will have to wait until Fall '95.
Modal-Inverse and Mixed-Inverse cases have been also checked out,
for both inviscid and viscous cases.
The streamtube-thickness variation modes are working as expected.
Aug 31, 1995
ISET has been found to have problems with low-solidity blades
which don't overlap (i.e. the stagger distance exceeds the
blade chord). This "uncovered" gap causes the streamwise
grid node allotment logic to fail. A simple fix has been
implemented.
Modifications for v 2.2
The AGS transition correlation Rts(H,Tu) has been re-tuned to better
match the original correlation data. The ISMOM=4 option has been
made somewhat more reliable. IPLOT has been cleaned up significantly,
with much nicer streamtube plots, and a new inviscid wake profile plot
menu.
Modifications for v 2.3
a) The MASS (total mass flow) global variable has been replaced with
the MINL (inlet Mach number) global variable. This makes the
code much better behaved for cases with inlet Mach near unity.
b) The dissipation coefficient function mu(M,Mcrit) has been modified
slightly to give better shock-propagation behavior.
c) A "polar" driver program POLAR has been added. This is essentially
ISES with a DO-loop wrapped around it. The DO-loop reads in a
specified operating parameter (Mach, P2/Po, etc.) from a spec.xxx file,
converges the solution for that parameter, and writes out the results
to a summary file and a surface-quantity dump file. It can therefore
conveniently generate a loss curve or operating line without
laboriously running ISES manually for each point. PPLOT and PXPLOT
for plotting the polar output files are not available yet.
Maybe for v 2.4, 2.5, 2.6 ...?
d) ISET now allows the inlet slope to be specified via a leading-edge
Kutta condition, which is an easy way to get a well-behaved grid
near small-radius leading edges.
e) Prescribed losses can now be input via the loss.xxx file.
This contains a ln(Po) vs m' function which is superimposed on
the flow equations. In effect, this is a nonzero prescribed entropy
function. See the User's Guide for more info.
f) Supersonic-axial inlet and outlet boundary conditions have
been reformulated. These actually appear to work now.
g) The Supersonic/subsonic-axial absorbing inlet boundary conditions
have been reformulated to permit nonuniform r(m') and b(m') distributions
over the inlet part of the grid.
Modifications for v 2.4 (unofficial release)
a) An arbitrary geometry definition capability is now supported.
A blade shape can now be defined via a bparm.xxx file instead of
blade.xxx. The bparm.xxx file contains an arbitrary set of
geometric parameters which are interpreted by a few black-box
user-supplied routines BPREAD, BPWRIT, BLDGEN. MISES doesn't
care what these routines do internally --- it just uses them
to determine the blade m',theta coordinates from the parameters.
b) There is a new Parametric-Inverse option which allows MISES
to drive the user-defined geometry parameters to obtain a best fit
to a specified target pressure distribution. This allows MISES
to be used as an inverse-design engine with ANY geometry-definition
system. This flexibility can be further enhanced by providing
a SUBROUTINE BPCON which allows imposing geometric constraints
(like area, thickness, bending inertia, etc.) on the
Parametric-Inverse calculation.
c) An arbitrary number of streamtube thickness b(m') variation modes
is now supported, although the number is still dimensioned to 2 via
the NBVRX parameter in STATE.INC. This can be easily changed if desired.
However, the format of the ises.xxx input file will also need to be
slightly modified to accept more than two modes. A more general means
to use these modes will likely be provided in future versions.
d) BLDSET can now read a blade shape in one of four ways:
i) x,y coordinates from a blade.xxx file (as before)
ii) x,y coordinates from an idat.xxx file (new)
iii) geometry parameters from an idat.xxx file (new)
iv) geometry parameters from a bparm.xxx file (new)
It can write out either the usual blade.xxx file or a bparm.xxx file,
the latter being possible only if options iii) or iv) were used
to in the parameters.
e) Color has been added to more plots to improve visualization.
Addendum (2/11/97 MD)
f) A new solver routine is now the default (the "old" solver is
ises/src/solve.old). The new one uses the black-box routines
LUDCMP,BAKSUB to perform the block-row inversion operations,
instead of the usual inline code. Using these routines has been
found to give more than a 2x speedup of the MSES airfoil code solver,
but here it gives only about a 15% speedup -- disappointing, but
I'll take it. No idea whether this new solver has better or worse
numerical roundoff behavior.
Modifications for v 2.53 (12/5/98 MD)
The MISES v 2.5 series have a number of very significant modifications
from 2.4 and earlier versions.
* BL Suction can now be imposed in a number of separate "slots"
* Wall cooling/heating can be specified. Previously,
only adiabatic walls could be represented.
* The boundary layer formulation has been significantly revamped,
primarily because the previous formulation was not very well
suited to the non-adiabatic case. The basic profile parameter
is now Hb rather than Hk. The profile and correlation
parameterization is described in document doc/blform.ps
* The e^n and AGS transition criteria have been revised.
The freestream turbulence level Tu, or the implied Ncrit factor,
are now referenced to the inlet velocity V1, rather than the
local velocity Ue(s). The standard methods, which used Ue,
imply that the freestream turbulence rms velocities are
proportional to Ue, which makes no physical sense. With the
new formulation they are assumed to be constant. The local
n(s) variable is still referenced to the local Ue(s), though,
so that n(s) will decrease as Ue(s) increases simply from the
changing reference value. Transition is therefore delayed
somewhat from the previous results when the flow accelerates,
although only intense acceleration like in turbine blading
will produce a noticable effect. The modified formulation
is described in document doc/ags2.ps
* The handling of the built-in geometry perturbation modes has
been cleaned up and generalized. The modes are now defined
in a modes.xxx file, rather than being hard-wired. The mode
shapes can be plotted in EDP.
* Overall translation,scaling,rotation modes can now be defined
for each blade. This nicely complements the geometry perturbation
modes which cannot move the overall blade.
* Constraints on the inlet and outlet tangential velocities can
now be imposed. Imposing the inlet tangential velocity but
not the total Mach number admits the unique-incidence condition.
Modifications for v 2.56 (20 Aug 04 MD)
* Fixed a number of linearization bugs in the suction implementation
Modifications for v 2.57 (7 Dec 04 MD)
* Changed the way M and Cp axis plot limits are set and changed in IPLOT
* Fixed surface node spacing in ISET so that symm airfoils have symm spacing
* Added Xshock constraint 21, useful for running flaky transonic polar sweeps
* Added Xshock constraint 21 to the list of constraints recognized by POLAR
* Added Xshock,Mshock inputs (optional) to 5th line of ises.xxx file
* Added a few additional parameters to the polar output file
Modifications for v 2.58 (20 Mar 05 MD)
* Added commands to IPLOT for dumping flowfield or BL to text files.
* Removed most of wall heat flux influence in Ctau_EQ expression.
This gave unrealistically large Ctau_EQ for very cold walls,
and unrealistically small Ctau for very hot walls. The modified
expression gives more realistic numbers for non-adiabatic cases.
The change has no effect for adiabatic cases.
Modifications for v 2.59 (22 Dec 06 MD)
* Fixed a few OOB errors (crashed with some compilers)
* Slightly extended some of the plot menus
* Changed the default blade side plot colors to blue and red.
The yellow sometimes didn't show up well on white backgrounds.
* Updated User's Guide
Modifications for v 2.60 (15 Jul 07 MD)
* The text-format read routines can now detect tab characters, which
are treated as spaces. Previously, a tab would cause a read error.
Modifications for v 2.61 (5 Nov 07 MD)
* Fixed a serious equation-indexing bug in ISES which showed up
only for cases with
a) more than one blade (e.g. with a splitter), AND
b) a shifted grid in the exit, AND
c) a negative exit slope.
Any case which didn't meet conditions a,b,c was completely unaffected.
I've never before run such a combinations of conditions, which is why
I never saw the problem.
Modifications for v 2.62 (12 Dec 07 MD)
* Fixed stagnation point displacement thickness updating.
No effect, except occasionally it would mess up the normal-offset
vector smoothing routine (DSTELL).
* Fixed stagnation point setup in ISET, which would sometimes end up
just inside the blade for blades with extremely small LE radii.
Modifications for v 2.63 (30 Jan 08 MD)
* Fixed ISET, which would sometimes fail the grid initialization
for extreme-turning blade shapes having sharp trailing edges.
* Added output of text file sens.xxx, which contains sensitivities
of various global variables (S1,S2,p2,omega... etc), with respect to
global specified parameters (S1spec, p2spec, geometry modes, etc).
This is in effect a short version of the unformatted dump file sensx.xxx,
which also contains sensitivities of the surface pressures, Hk,
and geometry.
* Updated and cleaned up the sensitivity-calculation procedure in
the LSinv.ps document. Also updated the main mises.ps document.
Modifications for v 2.64 (15 Jan 09 MD)
* Added re-read/re-plot commands (item 99) to the Surface-plot
and Contour-plot menus in IPLOT. These are useful when monitoring
iteration or optimization progress, since you can replot a new solution
without needing to quit and then restart IPLOT.
Modifications for v 2.65 (2 Nov 09 MD)
* Added Stanton number Ch(x) plot to surface plot menu.
This is simply overlaid on the Cf(x) plot as a dashed line.