Skip to main content



Nuclear Power Plant Simulation Session


Experience with Simulation of Nuclear Systems on Parallel Processing Computer Systems

Computer systems with parallel-processing capabilities are increasingly being used in technical and scientific applications. Very crudely, two types of computer architecture may be distinguished:
  • systems consisting of several (usually 2–8) high performance computers with a shared memory. No physical exchange of data between memory areas allocated to the individual computer is necessary
  • systems consisting of many (usually >16) computers which perform the calculations using their own distributed memories. Physical exchange of data between the local memories is required in such systems.
U. Graf, W. F. Werner

A Pump Model for Use in Real-Time Simulation

A fast running centrifugal pump model was developed and tested for use in a real-time simulator. Non-linearities due to the pump head and flow equations are treated by interpolation in a previously computed table. A low order integration scheme was enough to yield a converged solution at comfortably larger time steps. Agreement with experimental data is also discussed in this paper.
A. C. Barroso, L. V. Loureiro

Hydraulic Network Modelling for Real-Time Power Plant Simulation with Computer Aided Code Generation

The simulation of the power plant process for fossil and nuclear power plant training simulators covers a lot of hydraulic networks. For each flow path and pressure node of these networks non-steady state mass flows and pressures have to be calculated under real time constrains. The standard KAE procedure for this task applies the momentum balance to each flow path and simultaneously solves this system of equations for pressures and flows by means of matrix operations. In order to facilitate the handling of this standard procedure the software tool DECO (DEsign and Coding) has been developed. After inputting the topology of the hydraulic network, DECO automatically optimises the matrix in order to minimise the number of mathematical operations and additionally DECO generates Fortran source code to form up the matrix and to solve this matrix for pressures and flows.
R. Bakker, E. Große-Dunker, P. Leishman

Software Tools Session


SIPA, a Training and Engineering Simulator and AGLAE, a Flexible Workshop for Model Generation

An advanced nuclear plant simulator called SIPA (for “Simulator for Post Accidental Situation”) is under development for the French utility EDF. The Department of technical support to the Safety Authority (CEA — IPSN) intends to acquire a similar simulator as well for safety analyses.
The aims of SIPA are two-fold:
  • to train Shift Safety Advisors (SSA) to identify the plant situation and mitigate accident consequences
  • to be an experimentation tool for EDF design engineers.
The latter includes many analyses for present and future types of plants. These analyses are presently carried out by means of batch codes which are neither interactive, nor user friendly and which require manual programming.
Therefore there was an obvious need for tools for:
  • automatic generation of source code
  • creation of interconnections between models
  • creation of links between simulation code and interactive control images displayed on screens.
These tools are fully integrated in a hierarchical structure called “AGLAE workshop” which is presented hereafter and whose salient features are described.
Although this system has been originally developed for nuclear applications it could be used for the simulation of any type of industrial process.
C. Mathey, A. Valembois

An Environment for Parallel Structuring of Fortran Programs

This paper describes and illustrates the application of an environment for parallel structuring of Fortran programs. The characteristics of this environment are:
It uses a graphically displayed, hierarchical dependence graph representation of parallel programs.
It utilizes the module by module output of an optimizing compiler and a set of standard algorithms for interprocedural information propagation to generate and store both a hierarchical, global dependence graph for the Fortran program and a database of information, which can be used by the programmer/analyst to determine effective parallel program structures.
It generates default parallel program structures, which may be used directly or as a starting point for further analysis.
It automatically generates the code to implement and measure the parallel program structure selected by the user.
It provides a simulation modeling capability, which allows users to evaluate the execution properties of a given parallel program structure across a spectrum of execution environments.
It allows user input of dependence information to resolve the ambiguities that are always present in. Fortran programs and limit the effectiveness of automatic parallelization.
This paper has been structured so that the system’s capabilities are expressed through an example application involving the detection and expression of parallel program structures. Experience with this environment indicates that interactive support systems, which assist analysts in the generation of macro-level parallel program structures, can play a major role in the problem of restructuring existing Fortran programs to have efficient parallel structures.
K. Sridharan, M. McShea, C. Denton, B. Eventoff, J. C. Browne, P. Newton, M. Ellis, D. Grossbard, T. Wise, D. Clemmer

Advanced Modular Simulation Techniques for Nuclear Power Plants

An advanced technique to simulate nuclear power plants and other complex system was developed. The simulation technique is based on the DSNP (Dynamic Simulator for Nuclear Power-plants) simulation language. The DSNP is a modular modeling technique having several libraries with a large selection of component modules. Several examples are presented on how these modules are brought together to create a simulation program for a specific PWR, IITGR, and LMFBR plants. Results of some transients and accidents simulation are presented to demonstrate the DSNP capability and its versatility.
D. Saphier

Practical Experience with Software Tools to Assess and Improve the Quality of Existing Nuclear Analysis and Safety Codes

Large nuclear analysis and safety computer codes written more than five years ago pose a unique problem from the standpoint of Software Quality Assurance (SQA). Most SQA techniques designed to manage the life cycle of new software development do not adequately address the risk factors of software developed before these methodologies were well-defined or practiced. By defining the risks as related to factors such as correctness, reliability, maintainability, or portability, existing tools are used to assess these risks and provide an indication of the quality of the software.
Within the constraints of schedule and budget, we have applied software tools and techniques to existing FORTRAN codes determining software quality metrics and improving the code quality. Specifically discussed are INEL experiences in applying pretty printers, cross-reference analyzers, complexity analyzers, coverage analyzers, performance analyzers, and computer aided software engineering (CASE) tools and techniques. These have provided management with measures of the risk potential for individual program modules so that rational decisions can be made on resource allocation. Selected program modules have been modified to reduce the complexity, achieve higher functional independence, and improve the code vectorization.
N. H. Marshall, E. S. Marwil, S. D. Matthews, B. J. Stacey

Nuclear Core and Power Plant Simulation on High Performance Parallel Computer Systems

The present paper describes the development of a coupled neutron kinetics — thermal hydraulics program system for the calculation of steady-state and transient conditions in the core and for the modeling of the coolant loops of light water reactors. The resulting parallel simulation program 3D-SIM is designed to run on the SUPRENUM multiprocessor system which has been accomplished as a German supercomputer project during the last few years.
R. Müller, R. Böer, H. Finnemann

Artificial Intelligence Session


Knowledge-Based Systems to Support Dynamic Process Simulation

This paper summarizes major aspects of supporting systems dynamics modelling and simulation by knowledge-based systems or expert systems, respectively. Combinations of numerical simulation and symbolic knowledge processing techniques have proved to be very useful for improving the adequate representation of system knowledge, the efficiency, flexibility and consistency of models and the support and advice of users in system modelling. Based on actual requirements concerning the synthesis and experimental application of symbolic models, basic architectures and features of knowledge-based modelling environments are discussed. Results of prototype implementations are presented supporting the hierarchical description of discrete systems and the model synthesis in the domain of computer systems.
Axel Lehmann

Computational Fluid Dynamics Session


Two Phase Flow Analysis Capability of Advanced Computer Codes

Much effort was put into the development of computer codes for analysis of accidents in nuclear power plants in the last few years with interest focussed on the modeling of two-phase phenomena /1, 2, 3/. ABB Reaktor has investigated to what extent conventional phase-separation models are able to treat special two phase problems of a feedwater system. In comparison to this the characteristics of modern solutions with six equation models were explored.
W. Kohler, M. Schindler

Finite Element Weak Statement CFD Algorithms for Fluid-Thermal System Analysis

The weak statement theory is developed for construction of approximate solutions to laminar and turbulent Navier-Stokes (NS) equations for incompressible fluidthermal flow applications. Introduction of a domain discretization assists in finite element compact-support trial space basis construction. Implementation steps are highlighted for several NS reconstructions to account for the incompressibility constraint. Theoretical issues of accuracy and convergence are highlighted, and numerical benchmark test results document the important performance comparisons.
A. J. Baker

Neutron Kinetics Session


Real-Time Advanced Nuclear Reactor Core Model

This paper describes a multi-nodal advanced nuclear reactor core model, developed and benchmarked by CAE Electronics Ltd in collaboration with Koclas Logic Systems Ltd. (KLSL). The model is based on application of modern equivalence theory to the solution of neutron diffusion equation in real time employing the finite differences method. The use of equivalence theory allows the application of the finite differences method to cores divided into hundreds of nodes, as opposed to the much finer divisions (in the order of ten thousands of nodes) where the unmodified method is currently applied. As a result the model can be used for modelling of the core kinetics for real time full scope training simulators.
Results of benchmarks , included in this paper, validate the basic assumptions of the model and its applicability to real-time simulation.
Jean Koclas, F. Friedman, C. Paquette, P. Vivier

The Validity of the Point Kinetics Model During Reactor Start-Up

A variational principle made stationary by the time-dependent, continuous-energy, P-1 equation is used to derive few-group diffusion equations. Flux adjoint spectrum weighted few-group parameters result. However, numerical tests indicate that the standard flux-weighted parameters provide sufficient accuracy for most cases. A two-group, three-dimensional nodal method is then used to test the accuracy of the point kinetics model with reactivity computed by perturbation theory for a start-up transient, starting from cold shutdown with a source present, to full power. As expected, the point model using fixed weight and shape functions to compute reactivity (perturbation theory) provides completely erroneous results.
F. A. Tarantino, R. P. Jacqmin, A. F. Henry

Axial Flux Difference on Beijing Nuclear Power Plant Simulator

This paper describes the importance of the operational parameter Axial Flux Difference (AFD), gives the axial power distribution in different core lifetimes, and shows the difference of AFD between the beginning of lifetime (BOL) and the end of lifetime (EOL).
Zheng Fuyu, Luo Jingyu

Plant Analyser Session


APROS Nuclear Plant Analyser

Modular Plant Analyser of Loviisa Nuclear Power Plant is being developed by Imatran Voima Oy (IVO) and Technical Research Centre of Finland (VTT). The product will be ready in 1991. The work is based on the APROS Advanced Process Simulator that was developed by IVO and VTT during the years 1986–1988. The present paper describes the build-up of the Loviisa plant primary circuit model using graphical user interface and generic components. The secondary circuit model of Loviisa is constructed in the same manner. The entire power plant model thus obtained is used for the calculation of two example transients. These examples originate from the Loviisa 2 unit dynamical tests in 1980. The Modular Plant Analyser results are compared with the Loviisa Unit 2 measurement data. This comparison indicates good agreement with the data. The present work has been performed using the Alliant FX/40 minisupercomputer. With this computer the Loviisa model fulfills at present the real-time requirement with 0.5 second timestep.
E. K. Puska, M. Saarinen, K. Porkholm

Computer Based Education Session


Pressurised Water Reactor Simulation in the Training Environment

In March 1989, the Department of Nuclear Science and Technology (DNST) commissioned a digitally driven Display Array Simulation System for use in the training of personnel employed by the Ministry of Defence who are about to enter the Royal Navy’s Nuclear Propulsion Programme.
The paper gives a brief history of Pressurised Water Reactor Simulation within the DNST and an outline of the training courses leading to the requirement for the Display Array Simulation System. Focus is then placed upon the flexible use of real time simulation in the teaching of plant dynamics by the use of model generated data.
The use of interactive consoles and a large scale colour graphic display has led to the success of the Display Array Simulation System within the DNST. Realisation of the potential of the system has led to many other proposed uses for the installed system and the paper concludes by discussing some of these.
A. G. Wills

A Real-Time Core Model for Nuclear Power Plant Simulators

The STK (Space and Time Kinetics) core model was developed at Link-Miles Simulation Corporation for real-time simulation of LWR (Light Water Reactor) core neutronic behavior in normal operating conditions including start-up and shutdown, and transient events. Accuracy, stability, and speed are optimized in the methodology of the STK core model. The STK core model has been implemented so far with the Link-Miles developed RETACTtm [1] model in both PWR (Pressurized Water Reactor) and BWR (Boiling Water Reactor) full-scope training simulators, e.g. American Electric Power’s D. C. Cook, Tennessee Valley Authority’s Watts Bar, and SCE San Onofre PWR simulators, and KSU F1 and TVO BWR simulators. Moreover, The STK core model is currently being implemented for six BWR and seven PWR full-scope simulators.
Der-Yuan Lee, Jan-Åke Gyllander

Nuclear Waste Session


Ground Water Flow Analysis of Potential Low Level Radioactive Waste Disposal Sites Using Electrical Circuit Analogies

The analogy between electrical circuits and ground water flow systems is developed. The analogy previously required an extensive electrical network to obtain the desired results. This paper deals with adapting the analogy to a computer based electrical circuit analysis program. The application of the analogy is then demonstrated through a preliminary ground water modeling of a proposed Low-Level Radioactive Waste Disposal Site in Illinois, USA.
George H. Miley, Kevin J. Kuelske
Weitere Informationen