|
| |
January 13th, 2008
Kintecus V3.95 has been released. Kintecus V3.95 has a new
integrator (invoked via the -INT:4 switch) that
can handle low-temperature (300 K) combustion runs and some stiff chemical kinetic systems
that
the first two integrators occasionally cannot handle.
Kintecus V3.95 will not run at all without a valid unlocking key that is unique to each
computer.
Also, some minors bugs have been fixed.
May 27th, 2006
A new version of Kintecus has been released (Kintecus V3.90) and Atropos (V1.20)
Prime new features :
1) A new feature in Kintecus V3.9 is the ability to perform complex
hierarchical cluster analysis on
temporal concentration profiles of the network with/without experimentally obtained
temporal concentration
profiles. Hierarchical cluster analysis in Kintecus has the ability to group
related and unrelated parts of temporal
concentration profiles in a meaningful, quantitative way. This grouping allows a user to
clearly see patterns that
were initially indiscernible or hidden.
"Why should I care about that?"
Cluster Analysis can significantly help in answering questions:
>> Analyze ALL species and determine which species or groups of species (or
subgroups, etc.)
are positively, zero or negatively correlated to each other and with other groups/species
in
either a pictorially or numeric output or both. This allows one to answer questions such
as:
>> What concentrations of E and S cause EIS to positively (or negatively or zero)
correlate with
EI ? or with ES? Or both? etc. For combustion, you can now finally answer that
question that has been bugging you since you were seven years old: If I combinatorially
scan H2,
O2 and temperature over a wide range, do the O and OH radicals always positively correlate
and also positively correlate (or negatively or zero) correlate with other species?
>> Determine which species in one's experimentally determined concentration profile
are
positively, zero or negatively correlated with all the modeled species temporal profiles.
Again, this can be grouped into a pictorially or numeric output.
>> Which species’ concentration profiles tend to stay in the same range of
concentration values?
>> There are other questions one can answer utilizing the myriad of cluster
techniques Kintecus
provides that the author has not fully examined…
Here is a small sample from a cluster analysis on the simple enzyme inhibition model:
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
EIS************
*
8************************
* *
EI************ *
*
10************
* *
ES************************ * *
* * *
9************ *
* *
E************************ *
*
11***********
* *
I************************************************* *
*
12***********
* *
P************************************************************* *
*
13************
*
S*************************************************************************
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
A much more thorough examination can be read in the Kintecus documentation .
Cluster analysis becomes quite a powerful tool when coupled with combinatorial scanning
(see the Scanning Section
to see how to set up a Combinatorial Scan).
2) Several more reaction types
Why should I care about that?"
* Include those reactions that do not seem to follow
standard/enhanced Arrhenius or RRKM forms. This is
primarily for plasma and nuclear reactions.
3) The True Global Equilibrium feature is more robust in
solving systems near or at plasma temperatures and/or
multi-phasic systems.
"Why should I care about that?"
* When one was examining the true equilibrium composition of a system over a wide
temperature range, occasionally the system would not converge at some temperature values.
The final plot of such a system would show “spikes” at the non-converged values
and the user was forced to either drop that point, or change the non-linear solver values
by hand until the system converged. You DON’T have to do this anymore.
4) Coupling of combinatorial scanning with output files such as system
thermodynamic output, rate fluxes, species fluxes, cluster analysis, Jacobian output,
eigenvector/eigenvalue output of the Jacobians.
"Why should I care about that?"
* Helps one to determine ALL the possible maximum and minimum concentrations,
temperatures, System Cp, H, G, viscosity, rates, fluxes that a system can have if one
varied species X and Y (or temperature or pressure) in all possibilities and combinations?
MAJOR BUG FIXES
1) If you tried to compute the eigenvalue and eigenvectors of the Jacobian matrix, this
would cause an instant crash.
This has been fix as of last year, but I only handed that version out to those that asked
about this bug.
2) You can now create more than 100 (up to 1,000) Normalized Sensitivity Analysis Matrix
files.
If you are planning to perform chemical mechanism reduction and elucidation with Atropos
make sure you download the updated Atropos program (V1.1).
Other general enhancements and minor issues resolved.
==============================================================================
ATROPOS 1.20 HAS BEEN RELEASED ********
Mainly several bug issues have been corrected as well as several enhancements such
as the ability to read up to 1,000 normalized sensitivity coefficient matrix
files.
==============================================================================
June 8th, 2005
Kintecus V3.8 has been released. In addition to the
changes described below
under the May 1st, 2005 description. Here are some more
features:
Gas Heating/Cooling Via Compression/Expanding
==============================================
Compression/Expanding a gas will cause heating/cooling
via dynamic pV^(gamma) where gamma is calculated
dynamically. The ideal value of 1.4 for gamma is NOT used,
but is calculated from weighted heat-capacities of all
the species present at the respective current temperature.
The heat-capacities are calculated from the 15 coefficient
thermodynamic database.
Cell Compartments
===================================
You can use fractional rate-stoichiometric coefficients
for cell compartments of different volumes. For example,
to describe the peroxynitrate (PNA) diffusion/convection
from the cell-nucleus to the cell-membrane where the cell
membrane is, say, 1/7 the volume of the nucleus and
one also knows the diffusion-rate-constants between the
nucleus and membrane (lets just say, D=.1, D'=.01) for
PNA, the two reactions can be written as (keep in mind
0.14286 is about 1/7):
0.1 , PNA{nucleus} ==> 7 PNA{membrane}
0.01, PNA{membrane} ==> 0.14286 PNA{nucleus}
General Additions
==========================================================
- the regression/optimizer is more robust.
- Two more random number distributions added to the
Uncertainty Analysis: Poisson and Gamma.
- the Kintecus-Excel Graphical Interface can now
load/plot normalized
sensitivity analysis matrices and scans:
May 1st, 2005
This was posted in the Yahoo Kintecus User's group back in March
2005, so
here it is again....
If anyone needs the below features right now, I can directly email you
Kintecus V3.8 with those new features:
Equilibrium Constant Constrained Rate Constant Fit
===================================
If you have determined the equilibrium constant of a
reaction then you can constrain the backward rate
constant against the equilibrium constant during a
fit. This is accomplished by specifying the forward
rate constant of a reaction multiplied by 1/Keq, i.e.
(assuming Keq is 3.0)
1.0e+8? , E + S ==> ES
k1*0.333333? , ES ==> E + S
This allows one to reduce the total number of rate
constants by half, so if you have 20 rate constants
to fit, but you determined the equilibrium constants
of all 10 reversible reactions, then there are only
10 rate constants to fit.
Low Temperature Combustion Runs
===================================
Kintecus V3.8 will be able to model low temperature
combustion runs. You can have the initial temperature
of the model (such as GRI-MECH3.0) to start in
the temperature range of 200K to 500K and the model
will run and not crash.
More Robust Integrator
===================================
Large temperature spikes, volume compressions
and large concentration changes in external
user perturbations that happen later in a simulation
with thermodynamics turned on would normally crash
or require one to start with a very small Accuracy
and/or a small Maximum Integration Time which would
dramatically slow down a run. Kintecus V3.8 can
detect these spikes and adjust to finish a
simulation.
Bootstrapping Standard Deviations
===================================
Kintecus V3.8 will allow one to bootstrap fitted
or optimized parameters (ie. k, Ea, m, initial
concentrations/temperature, third-body enhancements)
to determine their actual standard deviations. This
allows for much more accurate determinations of standard
deviations. The default method that is in Kintecus
V3.7 and practically all other simulation programs
is the asymptotic approximation which is known
to give incorrect values and even magnitudes, yet,
it is practice in many referred journals and
publications.
Scanning Parameters (also called Sweeping)
===================================
Kintecus V3.8 will allow one to scan any parameter(s)
(initial concentrations, temperature, residence times,
third-body enhancements, TROE factors, etc.)
in a variety of ways and stepsizes. Sequential,
parallel-stop, parallel repeat, combinatorial scans
(sweeps) of any parameter or parameters will all be easily
accomplished.
CSTR/PFR inlet temperature flow
===================================
You can now specify a different global temperature for all
inlets of a CSTR/PFR run.
January 17th, 2005
uncertainty (represented by either standard
deviation bars or confidence bands) is, most likely,
the most important and last procedure one
should do to validate a model. Turanyi et al.[28] has
shown that NOT performing such a procedure
and using only one-run time profiles is quite naïve
as it is very likely the
chemical kinetics/thermodynamics and NOT transport/convection
processes that are
controlling the concentration and temperature
profiles. A good analogy would be a hurricane
forecaster that only shows one single
possible path a hurricane can take with no uncertainty
bands or an average path! In addition, the
uncertainty analysis in Kintecus also gives one a
"smeared-out" sensitivity analysis
on ALL the parameters/constants in a model as well as
possible maximum and minimum time profiles
concentrations (and temperature) can take on
during the course of all the simulations.
 | It is HIGHLY RECOMMENDED that if you are using the Excel
graphical interface, then
you should copy all your model, species, parm spreadsheets into the newer
Kintecus V3.7 spreadsheets. The newer spreadsheets have additional code to
plot uncertainty/confidence bands, system properties and various bug fixes. |
| Click on the Graphical Interface button on the
left to see various new types
of plots available! |
December 13th, 2004
| Kintecus V3.7 will be released soon... |
| Kintecus V3.7 will have a substantial overhaul of its
fitting/optimization. Kintecus
V3.7 will now output many more statistical descriptions about how well
the
fitted data compares with the experimental data, descriptions on the
uncertainty
of the selected parameters (such as rate constants, initial
concentrations, third
body enhancements, TROE factors, etc.) and statistical descriptions on
the
important residuals. In fact, you could say Kintecus V3.7 outputs many
more statistical gauges than other "SASSIER" programs that
just do statistics.
Click here for a sample enzyme rate
constants fit or here for the classic
Oregonator oscillator/clock reaction
constant fit! |
| Some commercial clients have been asking for an addition to
Kintecus that can
calculate uncertainty curves for the prediction. This is much like what hurricane
forecasters
do with hurricane paths. As one respected client states, "Can you imagine if a
hurricane
forecaster only showed the average path of a hurricane without the uncertainty band?
" Yes,
there would be bloody hell to pay. Kintecus V3.7 will incorporate this with an extremely
versatile means of specifying uncertainty in ANY of the constants (rate constants,
TROE factors, residence times, temperature, Cp, Cv, H, S, G, A, concentrations,
etc. etc.) in any distribution (uniform, gaussian, etc.) with any number of repeated runs
(which can range from several hundred to several thousands of runs
depending on the
size of the model). (see below)
 |
June 30th, 2004
- Kintecus V3.5 has been released! Please see the June
20th history below to see all the
additions, changes and bug fixes. There's one additional addition not listed below,
plotting
templates! For those tired of that old, drab looking concentration and temperature plots,
you have Kintecus use your own plot templates automatically!
June 20th, 2004
Kintecus V3.5 will be released soon, here is a list of the new
features
and bug fixes:
<< Dynamic Mode >>
- Now you can finally control Kintecus through the new dynamic
mode! Kintecus V3.5 has a new mode that allows one to pause Kintecus, examine the current
concentrations, temperature, rates, system properties, residence times, external
concentration, etc, alter any of those properties, and continue. This allows one to
control Kintecus dynamically in a feedback manner. Feedback control can be a very
complicated and broad area which is beyond the scope of this document. Please consult an
appropriate text book on this subject. Dynamic mode in Kintecus is invoked through
including the "-dynamic[:wait time]" switch on the command line. A very simple
example is provided through using VBA scripts to control Kintecus in dynamic mode is
provided as Excel Workbook, "Combustion_H2_Dynamic.xls". Again, any programming
language that can output files can be used to control Kintecus.
<< Optimization / Fitting / Maximization /
Minimization >>
- A new feature in Kintecus V3.5 is the ability to extract molar
extinction coefficients (also known as molar absorptivity or epsilon from A=epsilon * l *
c ) from a series of absorbance values or a series of concentration values multiplied by
some weight. A sample Excel Spreadsheet name Enzyme_Regression_Molar_Extinct.xls
demonstrates this feature. The fitdata worksheet now contains absorbance values (instead
of the usual concentration values) recorded at some wavelength. To extract molar
extinction coefficient(s) (or a weight factor) from your absorbance values, give the
-FITWEIGHT switch on the command line with these options:
-FITWEIGHT:X1:Y1[:X2:Y2:X3:Y3:Xn:Yn]
Where X represents the column number in your fitdata
file/worksheet and Y represents the starting weight guess.
- CONSTRAINTS! You can now apply a constraint to any parameter
you are optimizing/fitting. The constraints allow one to specify lower and upper bounds
for a parameter. Constraints are applied by appending your guess with (lower<upper)
where lower is the lower numeric bound and upper is the upper numeric bound. For example,
to constraint the initial temperature guess of "3000?" to lay between 1500 and
4000 simply type: "3000(1500<4000)?" .
- TWO (2) new optimization methods, Levenberg-Marquardt method
and "The Complex" (the Simplex method on steroids).
<< Heat Loss/Gain >>
- A new parameter field in Kintecus V3.5 is the External Heat
Source/Sink field. This field allows one to model External Heat Jackets or External
Temperature Baths or dynamic Heat sources and sinks. You can enter one of four types of
external heat source/temperature related items:
1) A numerical value representing an external Heat Source/Sink.
This value will be added to the dT/dt equation and with units in Kelvin/sec.
2) A filename containing a profile (formatted like a
"Constant File?" file) that represents a changing External Heat Source/Sink #
Profile (Filename) with units in Kelvin/sec.
3) TWO values: One for Conductance of the container/reactor,
followed by a COLON, ":", followed by the Extern. Temperature(K)
4) One value and a profile: the first number represents the
conductance of the container/reactor followed by a COLON, ":", followed by a
filename containing a profile (formatted like a "Constant File?" file) that
represents a changing External Temperature
<< Two new integrators >>
- The "classic" Gear BDF method
- The Adams-Moulton method
<< MORE EXCEL WORKBOOK SAMPLES >>
There will be a few more Excel workbook samples (though Excel is
not needed to run those Kintecus examples), such as computing a volume profile of a piston
and using that in a GRI-MECH 3.0 run.
************************
** MAJOR BUG FIXES *****
************************
There have been a number of minor bugs fixed and most of those
will go unnoticed, but if you are working with constant-volume (isochoric) adiabatic runs
(in kinetics mode, not equilibrium mode) or are applying a "Constant File" to a
species in thermodynamic mode then the following two major fixes will apply to you. Those
of you that only use Kintecus in Equilibrium Mode can ignore these two major bug fixes as
they only pertain to Kintecus in chemical kinetics mode.
<< MAJOR BUG CORRECTION 1 >>
* The constant-volume (isochoric) thermodynamics in some
non-isothermal adiabatic runs have been corrected so some users might see the correct
higher temperature. Again, this is model and initial condition dependent so some users
might not see any change while other might see final temperatures change upto 40% and
ignition delay changes of 10-30%. Example, the five GRI-MECH samples that come with
Kintecus show almost NO CHANGE, BUT the three ethanol combustion runs show some changes in
ignition delays. For those that are performing runs in CONSTANT-PRESSURE, then you can
safely ignore this bug fix, constant-pressure mode works fine.
<< MAJOR BUG CORRECTION 2 >>
* Running Kintecus in chemical kinetics mode with thermodynamics
and using a "Constant File" (an external concentration profile) on a species can
cause the previous versions of Kintecus to immediately bomb out with an
"Overflow" error. This made it appear it was an integration problem, but it
never was an integrator problem. The values in your Constant File were accidentally set to
either zero or to a colossal negative number. This has now been fixed and is working
properly.
May 31st, 2004
 | VBA Excel Bug: |
| If you store temperature/concentration/other
perturbation profiles within
an Excel Kintecus Spreadsheet or store thermodynamic databases or
Chemname files (by prefixing worksheets with "O_"), then Excel will
not write those files into the Kintecus Path. They will usually be written
to "C:\", but clicking on the RUN button
a SECOND time, forces Excel to
write to your Kintecus Path (usually "C:\Program Files\Kintecus" ).
*** FIX ***
You can wait until the next version of Kintecus to be released OR
simply
to remember after loading the Kintecus Spreadsheet, RUN the
simulation
at least ONCE whereby Kintecus will always write files to your Kintecus
Path and not "C:\" *OR*
modify
the VBA code simply by moving the following code in the Kintecus Run
Macro to right under the last DIM statement.
(Dim V1 As Double, V2 As Double, V3 As Double, nation As
Integer):
MOVE THESE PIECE OF CODE:
' goto kintecus directory
KPATH = Worksheets("CONTROL").Range("A8:A8")
If Right(KPATH, 1) <> "\" Then
KPATH = KPATH & "\"
End If
L = InStr(KPATH, ":")
PC = ""
If L > 1 Then PC = UCase$(Mid$(KPATH, L - 1, 1))
If L <= 1 Or PC = "" Or PC < "A" Or PC > "Z" Then
PC = "C"
ChDrive PC & ":"
ChDir KPATH
right under the VBA code:
Dim V1 As Double, V2 As Double, V3 As Double, nation As Integer
|
November 9th, 2003
| A new version of Kintecus is available: Kintecus V3.21.
This version corrects the bug
with the normalized sensitivity analysis feature. In addition, those
interested in
performing an Intrinsic Low Dimenionsal Manifold Analysis (ILDM) on a chemical
kinetic
system for use in 1D/2D/3D runs will definitely find the new eigenvalues/eigenvectors
and Jacobian output features helpful. |
| Be sure to check out Atropos
! This NEW KINTECUS SOFTWARE COMPANION
will reduce large, complicated mean old chemical kinetic systems into
puppy dogs and everyone loves puppy dogs.... |
|
