Head of Reactor Physics Department
phone (at work) | +386 (1) 5885-324 |
fax | +386 (1) 5885-377 |
e-mail address | andrej.trkov@ijs.si |
home page | http://www.rcp.ijs.si/~andrej/ |
CV |
Nuclear Data - Evaluation, Verification, Processing, Validation
CORD2 - PWR Reactor Core Design Package
GNOMER - Core Power Distribution with Thermohydraulic Feedbacks
DMR043 - Digital Reactivity Meter
LOADF - Power Reactor Core Monitoring
WLUP - WIMS-D Library Update Project
Running
Mountineering
Cycling
Rowing
An important part of my research activities are devoted to
nuclear data. Early work involved the implementation of the
FEDGROUP-C code for generating multigroup constants for
deterministic neutron transport calculations. Later, emphasis was
shifted to the NJOY code system (developed at Los Alamos National
Laboratory). I contributed to the WIMSR module for generating
constants in the WIMS-D Library Format and numerous other small
patches.
The success of a reliable nuclear data file depends on file
verification and validation. For this purpose the ENDVER package
was developed, which allows comparison of the contents of
evaluated nuclear data files with the experimental data in the
EXFOR database. The data retrieval engine was my contribution to
the ENDVER package (available from the IAEA), which complements
the database management and the Graphics User Interface prepared
by Viktor Zerkin. Together these components are part of the ENDF
data retrieval interface at the IAEA and the NNDC.
Data processing is the first step in data validation, which is
nowadays a standard step in the nuclear data evaluation process.
Most commonly this is done by preparing a library for a Monte
Carlo transport code like MCNP and simulating integral benchmarks
such as compiled in the ICSBEP and SINBAD compilations.
All of the above requires thourough knowledge of the ENDF format,
which is used worldwide for storage and exchange of nuclear data.
I am the co-editor of the ENDF-6 Formats Manual.
With the experience outlined above I can contribute to the
development of the EMPIRE code system for nuclear model
calculations, which is one of the major codes for nuclear data
evaluation. My main responsibility is data file assembly, and
formatting.
Model caclculations need to be combined with experimental data to
improve the performance of the evaluated data and to constrain the
uncertainties. To this end I worked with D.W. Muir, the author of
the GANDR System to use GANDR for nuclear data evaluation work.
The main achievement in the field of nuclear data is the
inclusion of several evaluations (thorium, protoactinium, isotopes
of tungsten, manganese) into the ENDF/B-VII.1 evaluated nuclear
data library. The work was done through broad international
collaboration under the IAEA.
In the package, utilities are available for automatic preparation
of inputs for the well known, WIMS-D lattice code from AEA,
Winfrith. The non-commercial versions of the code are available
from the NEA Data Bank.
Reactor core reactivity at constant core configuration can in principle be determined by measuring the doubling time of the neutron flux. This method is limited to small, positive reactivity values. It is too time consuming for practical applications in power reactors.
A better way of measuring reactivity is by solving the inverse point kinetics equations, using the neutron flux signal from the neutron detectors as input. Early reactivity computers were analogue devices. Their main disadvantage is that they can not model all features of the neutron point kinetics equations and that they need very careful calibration to work correctly. More modern digital devices are more robust, accurate and easy to use.
In the early 80's the first digital reactivity computer DMR042 was designed at the "Jozef Stefan" institute by Glumac and Vidmar. It was based on a microcomputer and used successfully at the Krsko NPP during startup tests before cycles 2, 3, 4 and 5. In the meantime a new digital reactivity DMR043 was designed, based on a PC computer and utilising better numerical algorithms and allowing the full non-homogeneous form of the point kinetics equations to be modelled. It has been used at the Krsko NPP ever since (the Krsko plant is now in its 16-th cycle of operation).
As an offspin of the high performance reactivity computer, a new method for measuring control rod worth was designed. It was tested during the startup tests in cycle 6 and used as the primary control rod worth measuring technique since cycle 7 in 1987. Its main advantage is the short measuring time. All it takes is the time for the measured control rod to be fully inserted into the core and to bring the reactor back to the original power leve. The savings in time (and money) are quite substantial, considering that one day of full power operation is worth between 0.5 and 1 million US dollars. For illustration, the startup tests before cycle 16 took just about 12 hours to complete!
The Krsko NPP is not the only plant in which the new control rod worth measuring technique is used. A publication by Westinghouse appeared a few years ago, outlining the "Dynamic Rod Worth" measuring technique. This is an idependent development, but reference is made to some of our earlier work. This method is now offered commercially by Westinghouse.
For the DMR-043 I contributed the software and several design
features, which make the device easy to use. I also made a major
contribution to the new control rod worth measuring technique,
particularly the error analysis and the implementation of the
correction factors, which provide sufficient accuracy of the
method for practical purposes.
The LOADF package runs on the process computer of the Krsko NPP. It contains GNOMER as the main calculational module. It solves the 3D neutron diffusion equation in cartesian geometry assuming core octant symmetry. It reads the necessary parameters from the Process Information System database. The calculated parameters are also stored in this database to be available to the operators as necessary. The main parameters displayed at present are the Xenon and Iodine distributions, axial power distribution, shutdown margin and various reactivity parameters.
The main design problems were related to the reliability of the information in the database and the relatively modest computational capacity of the process computer, which is about 10 times slower than a 200 MHz Pentium PC.
The LOADF package was delivered to Krsko NPP in April 1999. The
performance up to date is quite satisfactory.
Within the scope of the WIMS-D Library Update Project (WLUP), the NJOY code was made operational to generate data for the WIMS-D library. A number of updates were proposed, which are compatible with NJOY99 .
This is something I discovered in recent years. Much of the
credit goes to the trainer Urban Praprotnik with his positive
attitude and a few hints to improve my technique, which made
running painless and really enjoyable relaxation. I managed the
Vienna Marathon in 2011 in less than four hours. Due to
circumstances I did not make the full running trip from Ljubljana
to the seaside, but I made two halves (in 2010 and 2012). Maybe
next year...
There is some tradition in the family: my son Mitja is rowing for
Ljubljanica.
(This home page is under construction)
Andrej Trkov, Created 21-May-99, updated 2011.