1. Application of Interval Approaches to Discrete-Event Simulation: 
Yuri Merkuryev

	Interval methods are traditionally used in situations when it is preferably to operate not 
with point estimators of unknown variables, but with intervals that include their precise 
meanings. For instance, these methods are often used in analytical modelling for estimation of 
model parameters and forecasting model outputs. The contribution discusses application of 
interval methods to solving the following problems in discrete-event simulation:
	1. Preliminary, rough and fast analytical analysis of the problem before its detailed 
consideration by simulation. For example, if simulation is aimed to find optimal meanings of 
some system variables, then the area of possible solutions could be defined through that stage, 
with the following specification of an optimal solution in this area by means of optimization of 
parameters of the simulation model.
	2. Optimization of parameters of simulation models by taking into account confidence 
intervals of performance functions. The traditional approach to optimization of simulation 
models deals with mean-value estimations of these functions. At the same time, there could be 
situations when the most pessimistic (or optimistic) solutions have to be considered (e.g., when 
maximizing the minimal profit from operation of the simulated company).
	3. Identification of coefficients of analytical expressions, linking parameters of simulation 
models with their performance variables, during the sensitivity analysis or metamodelling of 
that models. The interval approach provides validation of the designed metamodel during the 
identification process. 
	Sample implementations of discussed interval methods are provided.



2. Simulation Game  "WinPROST - Training of Production Control": 
Kurt Matyas

At the Department for Industrial Engineering at the Technical
University of Vienna an interactive computer-based simulation game,
called WinPROST was developed that includes the topics production
planning, production control, scheduling, material management and
cost calculation.

Participants receive the opportunity to transfer their theoretical
knowledge into a "real world" firm, represented by the model,
without the risk of financial disaster. Almost immediately they are
able to see the difference between theory of production control and
the situation in a company, where suppliers of materials and parts
are not reliable, machines brake down unexpectedly, and workers fall
ill. 

WinPROST is an interactive business simulation game where
participant-groups (each 4-6 member group represents a company) have
to get orders for products that are not homogenous, in competition
with the other companies. They have to determine the amount of
materials and components needed and schedule orders to fill these
needs. Capacities that will be required to produce one of the
products must be determined in order to find out whether the
schedule is feasible or adjustments should be made by buying or
selling capacity units. 

Since the official start it is one of the most successful courses of
our department. Students get the experience in team-work and overall
understanding of a company's operations is created.



3. Model of competitive finance-industrial association: A. A. Naumov, 
M. L. Belozerov

     In work the approaches to formalization and are  investigated  to
modeling  of activity of finance-industrial groups( F.I.G.).In the ba-
sis a vehicle econometrics models of merge of industrial and financial
structures  is  fixed.  The feature of such models consists in block a
kind of their structure with inherent to them(him) "own" and "binding"
block by elements. The identification of parameters of model is execu-
ted on a method 3-íSE. Putting problem of activity and optimum manage-
ment ( control ) of finance-industrial association is investigated. In
the basis of target criterion of a problem of management  a  parameter
is fixed of competitive advantage determined as a measure in the space
of the factors - characteristics of functioning F.I.G.Solved a problem
of  revealing and estimation of prospects of competitive advantages of
groups in dynamics(changes) them claster. A problem of organization of
such functioning F.I.G. is investigated, at which ability to innovati-
ons, promoting to deduction and strengthening(amplification) of compe-
titive advantages is acquired.




4. Second order recursive random number generator for computer 
simulation: Chiang Kao


     To generate random number (RN) of long period for large scale simulation
studies, the usual multiplicative congruential RN generator can be extended
to higher order.  A multiplicative congruential RN generator of order two
with prime modulus 2^31-1 attains a maximal period of (2^31-1)^2-1 when the
two multipliers (a1, a2) are chosen properly.  By fixing a2 at -742938285, a
multiplier recommended for first order generator in a previous study,
approximately 1.1 billion choices of a1 which are able to produce RNs of maxim period are 
investigated in this paper.  Via the spectral test of dimensions up
to six, 14 sets of multipliers (a1, a2) exhibit good lattice structure in a
global sense with a spectral measure greater than 0.84.  Ten of these
multipliers also pass a battery of tests for detecting departures from local
randomness and homogeneity.  Furthermore, the execution time is promising on
32-bit machines.  In sum, the second order generators devised in this paper
possess the properties of long period, randomness, homogeneity, repeatability,
portability, and efficiency, required for practical use.



5. Multiobjective optimization in connection with the active system 
theory: Arslanov M.Z.


        This report concerned the actual problem of using in the
active production system control the methods and models of active
system theory [1,2]. The important components in control mechanism
of such system are the planning procedure and stimulation system.
In the base model of active system theory this components are
investigated sufficiently fully. One of the main results is the
investigation of coordinated planning procedure, properties and
construction the coordinated plans set.
        This report represents the new results in the active system
theory in connection with multiobjective optimization.
        The first result is the proving that two problems are
equivalent: the problem of optimal coordinated planning and the
problem of multiobjective optimization.
        The second result based on the previous is the new
convolution of criterions in multiobjective optimization which is
generalization of well known Karlin's convolution and allows the
problem of construction the Slater-optimal points set to reduce
into the scalar optimization problem.
        The methodology of the active systems theory is applied
to designing the control mechanisms for enterprises of non-ferrous
metallurgy and machine-construction.

        REFERENCES

1.A.A.Ashimov et al.,Coordinated Control of Active Production
Systems, Moscow, Nauka, 1986 [in Russian].
2.V.N.Burkov, B.Danev, A.K.Enaleev, V.V.Kondrat'ev, T.B.Naneva and
A.V.Stchepkin, Large Systems: Simulation of Organizational
Mechanisms, Moscow, Nauka, 1989 [in Russian].



6. Multicriteria bargaining procedure for a discrete multistage 
system: Lukasik, Stanislaw


 An interactive negotiation procedure for a discrete multicriteria bargaining is considered.The 
approach corresponds to the theory of multicriteria bargaining 
developed by Wierzbicki 1988 and Bronisz , Krus 1993 , for the static,continuoussystems.In 
this paper we consider a multistage, discrete version of this problem.The model and the 
procedure refer to the planning of development of the interegional highway network.We 
consider a set of players, representing regions , each with several objectives, such as investment 
costs, enviromental impact, social and economic impact and the state strategic aspects.
The problem consists in supportig the players in reaching a Pareto nondominated 
solution,agreeable and close to their preferences.
This solution can be reached , when they informationaly cooperate with other.
The seqence of the negotiative iterations , leading to the solution, have been 
obtained by taking into account of individual greedy (discrete gradient) indexesand mutual 
cofidence indexes.As a rerence point we employ utopia point relative 
aspirations, whitch reflects particular preference of the players.
The interactive procedure provides the jointly specified improvement direction  in each of the 
iteration . 



7. Hypothesis testing and risk minimization in the statistical 
validation of simulation models: Nicholas A. Nechval, Konstantin N. 
Nechval , Alexander 0. Svarevsky


The problem of validating computer simulation models of real, observable systems remains 
today perhaps the most elusive of all the unresolved methodological problems associated with 
computer simulation techniques. A common form of objective analysis for validating simulation 
models is statistical hypothesis testing, where there are two important wrong decisions 
involved in testing the validity of a simulation model under a given experimental frame and for 
an acceptable range of accuracy: (1) Rejecting the validity of the model when it is actually 
valid, and (2) Accepting the validity of the model when it is actually invalid. The probabilities 
of making these two wrong decisions are the risks (the model builder's risk and the model 
user's risk) involved in the validation process. The purpose of this paper is to consider the 
problem of validating computer models in the light of contemporary thought in the fields of 
philosophy of science, economic theory, and statistics and to give a methodology for 
constructing the relationships among the risks, acceptable validity range, sample size and a 
threshold of test when statistical hypothesis testing is used for validating a simulation model of 
a real, observable system. The methodology is based on parameter-free statistics (whose 
distribution does not depend on the unknown parameters) obtained from the generalized 
maximum likelihood ratio and allows one to eliminate the unknown (nuisance) parameters from 
the problem. A numerical example is given to illustrate the suggested methodology.

Key Words and Phrases: validation, simulation, generalized maximum likelihood ratio, 
nuisance parameter, parameter-free statistic, hypothesis testing, risk minimization. 






8. General queueing insights for reengineering of administrative 
processes: Nico M. van Dijk


Reengineering and business process redesign are two concepts that reflect the aim in business 
environments to make operations more efficient and to optimize the use of capacities and 
skills. An important performance measure is the response or total process time for a 
production or job to get finished For manufacturing systems this measure has received 
considerable attention. For administrative processes in contrast. general insights and tools to 
actually reduce such times seem to be lacking. Two major remains are: 

  Task times in administrative processes are typically non-deterministic (or random).
  Different qualifications of employees and special requirements have to be taken into 
account.

This paper will show that queueing theory can he helpful. General insights will be presented 
and illustrated by a case-study.






9. Interval linear programming: Vitaly I. Levin

     Let's consider problem of Boolean linear programming
                  cx =max                               (1)
                  ax < b,                               (2)
                     =
            vector x has components 0 or 1              (3)
     In interval  statement, when c=[c1,c2], a=[a1,a2],b=[b1,b2].
Multiplication of  interval by number and addition of intervals in
(1), (2) fulful by general rules.  Decision of (1)-(3) demands  of
intervals comparasion  and  choose  the  greater  and  the  lesser
interval. This  comparation  we  can  introduse  analogically   to
number's comparasion. Here we  get  such  results:  1)  coensiding
intervals; 2) interval moving to the right-greater,  and moving to
the left-lesser;  3) if one interval covers another, then they are
incomparalle.
     Parallel with  (1)-(3)  let  introduce  two usual problems of
Boolean linear programming 1) lower boundary and 2) upper boundary.
Problem 1  we get from problem (1)-(3) by replacing of coefficient
c by lower boundaries c,  and problem 2 - by replacing c by  their
upper boundariec   c.   Limits'  systems  of  bouth  problems  are
identical and recived by replaycing in (2)  of  coefficients  a  a
first by  their  lower  boundaries,  then  by upper boundaries and
joing of all reciving inequalities in one system of  inequalities.
Let's call  two  general  problems  of  Boolean linear programming
coordinated if their decisions' sets have even  only  one  general
decision.
     Theorem 1.  Set of decisions of problem (1)-(3) is  equal  to
intersection of  decisions'  set of its lower and upper boundaries
problems.
     Theorem 2.  In  order that problem (1)-(3) had decision it is
necessary and sufficiently  that  its  lower  and  upper  boundary
problems were coordinated.
     So, in difference from geheral  problem  of  Boolean  lineare
programming interval problem not always has decision.



10. Matrix games with interval parameters: Vitaly I. Levin

     Let's consider game of two  persons  in  interval  statement,
when  pay matrix A=[aij],aij=[aij1,aij2],  i.e.  prize aij- of the
1-st person at his strategyiand strategy of the 2-nd person j  has
the form of interval. Let multiplication of interval by number and
adding of intervals at definition of pay  function  carry  out  by
usial rules.  Let introduce netion of interval game's decisions so
as it is introdused by fon Neuman for usual games.  Then decion of
interval  game  demands  of intervals'comparations and choosing of
greater and intervals.  Introducing this comparations analogically
to  number's  comparation we'll get:  1) coinciding interrvals are
equal;  2) interval moving to the right is greater and  moving  to
the left lesser;  3) if one interval covers another, then they are
incomparable.
     At the  same time let's consider two general games:  1) lower
boundary and 2) upper boundary.  We get the game 1  from  interval
game  by  changing  off prizes aij by their lower boundaries aij1,
and game 2 - by changing aij by their upper boundaries aij2. Let's
call two general games coordinated even if they have only one pair
of  decision  with  identical  for  bouth  games   optimal   mixed
players'strategies.
     Theorem 1.  Game with interval pay matrix has decision if and
only if its lower and upper boundary games are coordinated.
     Theorem 2.  Decision of game with interval pay matrix we  get
by  joint  of  cordinated  decisions  its lower (x,y,w1) and upper
(x,y,w2) boundary problems and has form (x,y,[w1,w2]).


11. Applications of block, process and object oriented language in 
simulation: Eugene Kindler

We used a method called fictive simulation (further FS) to implement software
products: the substance of FS is to describe a fictive dynamic system S in a
language suitable for simulation, so that S generates the same data as the
desired software tool should do. FS has been applied in Prague since 1970,
the last case concerns testing configurations of containers and generating
the shortest paths in container terminals.
     Often a result of FS could be used inside a routine of a true (not
fictive) simulation model or in a rutine governing a simulation study (i.e.
a sequence of simulation experiments). In such a case we meet nested simulat-
ion models, with different simulated times etc. In order to overpass the
obstacles, we used an object-oriented language with local classes and with
"life rules" declared for any class. The nesting of FS was a good experience
for it but then we implemented also nesting simulation models independent of
FS, e.g. those of systems containing simulating elements.
     The problems of nesting simulation models and their solution by means of
the object, process and block oriented languages will be explained and some
demontration programs for PC will be at disposal.


12. The Development and Applications of the Simulation System 
SITA: G.Jonins, J.Sedols

    SITA (Simulation and Iteration Algorithms) is a system of description
and simulation tools, developed in 
the Institute of Mathematics and Computer Science at the University of
Latvia, and it is designed for the investigation of the public utility
systems. SITA contains model description tools, which are remarkable for
their shortness and simplicity, and they are usable both for the discreet
event simulation and analytic modeling. In the case of analytic modeling
SITA automatically ensures the design of the equilibrity equations according
to the model description and their solution by iteration method.
    During the recent years (1992 - 1996) SITA is supplemented with several
new tools, which are considered and illustrated by examples in the full
paper. They are: the programming environment AMIS, the analytic modeling
system SITA/E and the dialogue tools.
    The environment AMIS contains the supplementary (service) programs,
which are common for all the sub-
systems of SITA. AMIS ensures the maintenance of the programs, their search,
execution in various modes, the graphical presentation and several other
often used actions.
    SITA/E represents the sub-system with the table-like model description
and analytic modeling facilities, which give out the exact numerical
results. The application for scientific research of SITA/E is illustrated by
two examples.
    The dialogue tools ensure the interaction of the model and it's user or
with some other program. They can be used to develop the business games, to
connect the animation tools, and for several other purposes.


13. ISDN-based Videoconferencing System: Fan Jiang, Hu Yun, Kari 
Kyttänen, Risto Lahdelma

Helsinki University of Technology

Keywords:	videoconferencing, telecommunications, ISDN, training,
business process re-engineering

This paper describes a project about videoconferencing via ISDN (Integrated 
Services Digital Network). ISDN employs all-digital transmission and
switching.  A very fast connection (call) set-up will be provided for local,
national and  international calls. The subscriber interface will have
sufficient capacity not only  to handle voice and video communications but
also data communications directly.  The services can operate concurrently if
desired. By using a single telephone line,  the user can make several phone
connections while still maintaining the computer  connection.

The system is based on Pentium/PC:s with Nokia Mediastation displays 
containing integrated video and audio equipment. The application software runs
 under the Windows 3.1 operating system. The telecommunication system is based
 on the H320 standard hardware and software supplied by Bitfield.

The system is flexible and requires minimal set-up time and effort for
establishing  a point-to-point connection. This makes it ideal for
multinational ad-hoc  videomeetings. This creates new opportunities for the
communication structure of  international distributed organisations.



14. Adaptive Computer Simulation of Numerical Algorithms: 
Arsenjev D.G., Ivanov V.M., Kul'chitsky O.Yu.


     In this report the adaptive control theory of stochastic
systems is used in conformably to the untraditional objects, as
Monte-Carlo calculating procedures of many-dimensional
integration and numerical decisions of of integral equations. The
control action for these objects is the function of density of
unit distribution of a random mesh of integration. The criterion
of an optimum of functioning was chosen by the criterion of
accuracy of calculations, which has analytical representation for
both considered objects of control.
     One of the determining features of statistical computing
procedures is their recurrence, witch makes them adequate to
dynamic systems, the control theory of which is well developed.
The application of the information control theory for objects of
this kind enables significant increases of the efficiency of
calculations.
     Usually the characteristics of algorithms of calculations
are chosen from the conditions of their convergence on a class of
integrable functions or integral equations, that does not
obviously take into account the individual properties of that
specific object of calculations, to which the algorithm is
applied. The idea of the report consists in application of
methods of the adaptive control theory for tuning and adaptation
of calculation algorithms during their functioning.
     Such approach appears effective thanking the opportunity of
receiving the analytical decisions for variational problems of
optimization of algorithm accuracy on function of density of unit
distribution of a integration mesh. Thus optimum density of unit
distribution proves to be essentially dependent from individual
properties of integrable functions and integral equations. This
dependence is the main for the application of methods of adaptive
control, enabling considerably to increase the efficiency of
computing processes.


15. Simulation based evaluation of control policies for inventory 
management in multi-echelon physical distribution systems: 
Vanmaele

Over the years, many different types of inventory control policies have been 
designed in order to optimize inventory location and management in 
multi-echelon physical distribution systems. Among them, we can identify 
well known policies such as continuous replenishment (based on reorder 
points), fair share allocation, distribution requirements planning, ...
To analyze the performance of these different policies, we designed a 
simulator to model basic multi-echelon structures, allowing to measure 
important performance indicators such as cost (transportation, holding and 
ordering cost) and service level. The main experimental factors included in 
the system are : demand variations, volume distribution over end points, 
resource capacity (production capacity, inventory capacity, transportation 
capacity), ... 
The experimental design used for the factor analysis is based on a Taguchi 
orthogonal array design.
In this paper we will present the results of our experimental runs and we 
will formulate some conclusions concerning the performance of different 
control policies under varying conditions.



16. On the proper choice of solution routines for combinatorial 
optimization problems: Burkard, Rainer Ernst

Abstract:
Hard combinatorial optimization problems can be divided into 
two classes: some problems have the property that optimum and 
worst objective value will diverge, if the problem size 
increases, whereas for other combinatorial optimization problems 
optimum and worst solution converge to a common value. Knowing 
this behaviour can simplify the choice of an adequate heuritic 
for gaining good suboptimal solutions.


17. Social simulations games and evolutionary mathematical models 
as tools for knowledge creation in business process development: 
Minna Forssen-Nyberg, Eeva Piispanen, Ulla Seppala, Riitta Smeds


Any organization that deals with a changing environment ought not only
to process information efficiently but also to create information and
knowledge. There are two dimensions of knowledge creation: tacit
knowledge and explicit knowledge. Tacit knowledge is personal, context
specific and therefore hard to formalize and communicate. Explicit
knowledge refers to knowledge that is transmittable in formal
systematic language. The organizational knowledge creation process is
a spiral process starting at the individual level and moving up trough
expanding networks of interaction that crosses sectional and
organizational boundaries. The main question is, how the knowledge
creation and accumulation process will be initiated, created and
managed, especially by means of social simulation games and
evolutionary models. 

The aim of this presentation is to define social simulation game as
tool for knowledge creation, training and development and describe to
how social simulation games and mathematical models could be used as
methods for creating a knowledge spiral in real organizational
contexts. We will discuss the following items: 
* social simulation games in externalization of tacit knowledge of employees 
* evolutionary models for visualizing the explicit knowledge of the
organization 
* possibilities to combine these approaches for knowledge
creation in business process development.


18. Goldratt's simple production game - what can we learn from it?: 
Rand, Graham

Goldratt, the developer of OPT and the Theory of Constraints, and his 
followers,  have presented a simple dice game to instill various important 
lessons about manufacturing.  The focus is on the effects of statistical 
fluctuations and dependent events.  Students usually enjoy playing the 
ame, but is it an effective mechanism for teaching lessons about manufacturing?


The title of my session(s) - I would like 2x 1.5 hour sessions if possible 
to allow the games to be played - is "Learning about manufacturing through Games"

I expect the following paper to be in my session(s) and am hoping for a 
paper from Jim Freeman in Manchester, England.

The In-Class Manufacturing Game: Salwa Ammar & Ronald Wright  
Department of Business Administration
Le Moyne College,  Syracuse,  NY 13214, USA
wright@palm.lemoyne.edu


        The In-Class Manufacturing Game is a tool developed to demonstrate the
various aspects of manufacturing and inventory planning and control.  It is
designed as a two or more group competitive game, where each group plans and 
implements a manufacturing schedule.  The game is a very effective
educational tool that brings together several topics of operations
management and is very well received by students at the MBA as well as
undergraduate introductory operations management level.
        In this presentation we will describe several different ways in 
which the game can be used and play a shortened version.


19. Operational Process Modeling and Simulation: Oscar Barros



When performing reengineering in an operational business process,
alternative redesign solutions may appear. In selecting the right redesign,
performance measures should be considered. However, it is difficult to get
values for these measures for a resign that only exists in paper. So a
formal process modeling approach that allows for the simulated execution of
a process is needed to get better estimates of process effectiveness and as
a mean to improve redesigns. 
We propose a business process simulation approach specially suited for
operational processes. It is based on a general process model and an object
oriented implementation of the general model components. This model posits
that any operational process can be represented by using the following
object classes: a physical class that models the actual operations -e.g.
machine processes over materials or administrative operations in a credit
aproval process-; a regulation class that make decisions on how to assign
the physical objects, based on the knowledge of their state; and an event
class which generates the events that drive the dynamics of the simulation
-e.g. new orders for products to be manufactured or new clients that request
credit.
We show the use of our approach in the modeling and simulation of a variety
of manufacturing and services processes and calculation of performance for
their redesigns.



20. Enhancing of simulation and planning tools by integration of 
human related factors : I. Ehrhardt, P. Lorenz, B. H. Müller

Presently manufacturing simulation and planning tools allow primarily statements about the 
economy and effectivity of the considered systems. Manpower is often only modelled as a 
resource, like machines or other equipment components. But in fact the person with his 
creativity and flexibility is one of the most important elements of production systems. Missing 
information about burdens and threats of the person leads to a loss of quality of the used 
models. Risks of health can jeopardise the economy of production systems, too, if the person is 
considered as a subsystem.
In this paper methods and results of the application of the simulation of prospective 
consideration of the interactions between the system components technology, organisation and 
personnel in the area of planning and control of production processes are presented. As part of 
the research project EMSIG (Ergonomics in Material flow Simulators InteGrated) now 
features and components were created for the inclusion of knowledge about manpower in 
technical planning systems. By assigning burden and stress information to processes, beside 
traditional result values of a simulation, like run times and capacities, new results may be 
generated characterising the situation of burden and demand of the envolved personnel. First 
applications showed that the target values "increase of the economy" and "reduction of 
physical burdens" do not have to contradict each other.  
Current work is dealing with a more complete consideration of relevant human related 
elements. Thereby burden, safety and danger relevant information are provided and prepared 
for scheduling and simulation methods.
In that way the mapping quality of the models of production and material flow systems is 
improved. Advancing scientific analysis and evaluation methods in the planning process can 
support the decision on variants of planned concept on a broader information basis. 
Insufficiencies and health risk for the staff can be recognised already on the schedule stage of a 
operation system and avoided through suitable steps.



21. The In-Class Manufacturing Game: Salwa Ammar & Ronald 
Wright

Department of Business Administration
Le Moyne College,  Syracuse,  NY 13214, USA
wright@palm.lemoyne.edu


        The In-Class Manufacturing Game is a tool developed to demonstrate
the various aspects of manufacturing and inventory planning and control.  It
is designed as a two or more group competitive game, where each group plans
and implements a manufacturing schedule.  The game is a very effective
educational tool that brings together several topics of operationsmanagement
 and is very well received by students at the MBA as well asundergraduate 
introductory operations management level.
        In this presentation we will describe several different ways in which
the game can be used and play a shortened version.



22. Modelling and simulation of a once-through steam generator for 
control purposes using matlab-simulink: Wlodzimierz Stanislawski


 An application of present-day software resources to analysis
and synthesis of multivariable control systems is presented in
the paper. Specifically, various dynamical models of a
once-through steam generator are analysed .
The matematical models of the steam generator are coded in
MATLAB-SIMULINK, which allows to simulate the dynamic behaviour
of the steam generator at any disturbances.
The obtained models are verified on a basis of operational data
from the BP-1150 boiler at the Opole Power Station, and
simulation data.
The elaborated models of the once-through steam generator as a
multivariable system can be used in analysis and synthesis of
control systems.
The work contributes to an important issue of adaption of
control systems for Polish power plants to automatic control of
power and frequency within the Polish Electric Power System as
well as its necessity to comply with the standards of the
European ElectricPower System.



23. Administrative Information System of the Government of the 
Republic  of Lithuania (Business Process Reengineering Approach): 
Evaldas Kulbokas, Darius Maikstenas, Eugenijus Telesius

The Specification phase of Administrative Information System
 of the Government of the Republic of Lithuania (VADIS) started
 in 1995 and is in progress now. Sekasoft Ltd. in cooperation
 with two Kaunas Universities have won the tender and are involved
 directly in the project. All properties of VADIS conceptual
 model focus on business processes design and reengineering
 methodology. This methodology is known as Flores-Winograd
 (ActionWorkflow) approach and recognizes that work is accomplished
 by people delivering on commitments, not through merely moving
 documents around an institution. We are not focused on traditional
 Information Systems Analysis by which just the efficiency of
 document flow within the institution is increased. The VADIS
 institutional process maps providing a description of the
 interactions between users were created using ActionWorkflow Analyst.
 Now we are only in the first stage of design and are evaluating
 ActionWorkfow System, FlowMark and Lotus Notes system as possible
 implementation tools. Instead of using computers merely to automate
 an inefficient paper-based processes, we proposed implementation
 of groupware infrastructure which will facilitate institutional
 processes reengineering.


24. Simulation prototyping: Ingolf Stahl



A simulation model is successful if it leads to policy action,
i.e. if it is implemented. Implementation studies show that in
order for a simulation model, i.e. a computer based, simplified
picture of a real system, to be implemented, it must have good
correspondence with the mental model of the system held by the
user, or "buyer", of the simulation model. Because of this, the
user must feel confident that the simulation model corresponds
to this mental model. Therefore the user should have a basic
understanding of how the model functions.

Models for implementation must hence be developed step by step,
starting with a very simple model with an easily understood
structure. After this simple model, the simulation prototype,
has been presented and explained to the user, a more  detailed
model can be developed on the basis of the feed-back obtained
from the user.

Suitable characteristics of a simulation package or language for
simulation prototyping are discussed, e.g. with regard to the
speed with which small models can be written and the ease with
which models and output can be explained. These types of
simulation systems are finally set in contrast to packages and
languages for the final, detailed simulation.


25. Simulation of a Flexible Manufacturing System for Cost 
Estimation: Steen Nielsen

The Aarhus School of Business and
the Stockholm School of Economics

A fairly new method in the field of cost accounting is the use of
discrete-event simulation for estimating the indirect costs of
producing a product in an advanced manufacturing system. The
paper first gives a brief overview of earlier examples of using
simulation in this area. None of these examples deals, however,
with the application of simulation to an advanced manufacturing
unit existing in reality.

The paper then reports on the IBM Jarfalla FMS/CIM-cell outside
of Stockholm, where two types of parts for computer printers are
manufactured. The present cost accounting system, where volume is
the only cost driver, is next presented. This system does not
take into consideration that time is a very important cost driver
and it also does not include the stochastic variations in arrival
of production orders and in machine times.

In order to handle these aspects, a new cost accounting model is
constructed, based on three types of cost drivers: time, volume
and value added. Since, in particular, time is a stochastic
variable, a stochastic discrete-event simulation program is
needed to do the calculations for the cost model. By a large
amount of  repeated runs, cost per unit can be established to lie
between two limits with 95 per  cent confidence. Simulation
generated diagrams of "cost surfaces", showing the average unit
cost for various quantity combinations of the two parts, present
a new aspect on the solution of the problem of joint cost. Cost
of unused capacity is also calculated.

The simulation program is done in micro-GPSS, a streamlined
easy-to-use version of GPSS, with special facilities for file
handling and statistic analysis of repeated runs, making it
suitable for this type of fairly small model.



26. Associative modelling and control of complex objects: 
L.Rastrigin, G.Vulf

Article concerns modelling and control of complex objects, which do
not have a formal description, i.e. they do not have the mathematical
model. Such objects can be of economic, social, psychological, etc.
nature. There is by no doubt need for modelling and control of such
objects, and the accuracy of modelling and control can be comparably
low, taking into account fuzzy nature of these objects. The initial
data is the observations (precedents) report of object's functioning.
Associative modelling and control of such object is based on the idea
of associative solutions, i.e. finding in the report solutions which
are in certain (associative) sense close to demanded and find the
solution on this basis. The algorithm of finding such solution must
have the ability of maximal restoration of the report. For this
purpose certain optimisation task is solved, which enables obtaining
the algorithm, adapted to initial report of the task. For the
associative solution of different type of tasks it is necessary to
change the initial report to the form that is characteristic for the
task of this type. The associative algorithm is universal and requires
only adaptation to the obtained report. Article describes base
algorithm of associative decision making using report of observations
(precedents) and applying it for solving the tasks: . modelling of
dynamic systems . control of dynamic systems . identification of
objectives of dynamic systems . forecasting the behaviour of dynamic
systems
Included are also the results from using the associative approach in
the examples of the model tasks of mentioned types, showing the
effectiveness and possibilities of the method. Proposed associative
approach allows solving the task without synthesis of the system's
model. Its effectiveness improves by increasing the volume of initial
report. Most effective associative approach is for the tasks, in which
relation between situation and solution is close to linear. In the
case of significant non-linearity the volume of the report should be
increased.


27. New IT Infrastructures - How can they contribute to improved 
management of your business: Eriksson, Inger, Nissen, Hans-Erik 
and Pispa, Juha

If the purpose of business is to make money for its owners, the
purpose of investing in information technology is to support this
business. However, somewhere on the way from a promising idea to an
actual implementation this truth is often lost. The paper argues
that managers, by making some critical distinctions and asking some
central business oriented questions, can improve the return on their
investments in information technology. The paper suggests when and
how this can be done using illustrations from areas of application
such as electronic data interchange, office information systems and
strategic information systems.


28. Optimisation of Control Variables for an Automated Press-Line 
using Simulation Technique and Factorial Design: Hans Bylesjö

Dept. Of Manufacturing Engineering
Luleå University of Technology
SWEDEN

ABSTRACT
Many variables influence the performance of a manufacturing system; some are controllable 
others are not.  It is often the case that the behaviour of the system is measured by a number of 
parameters; for example levels of work in progress, lead time, manning levels etc.  Optimising 
the system for one criteria is often at the expense of sub-optimisation of the others.  Thus, 
optimisation is a question of balance in order to achieve a system ‘response’ that is acceptable 
and matched to the strategic and operational goals of the company.  This article explores the 
use of a simultaneous optimisation method to find the balance between the most significant 
controllable factors in order to achieve an optimum system response.

A fully automated press-line at SCANIA´s Luleå plant was studied using simulation techniques 
and factorial design.  The results showed that it was possible to find the relationship between 
three controllable factors necessary to optimise the performance of the press-line to best match 
SCANIA’s strategic and operation goals.




29. Application of multivariate analysis in combination with discrete-
event simulation: Ulf Eriksson

Dep. of Manufacturing Engineering
Luleå University of Technology
SWEDEN

Simulation techniques are commonly used to model manufacturing systems in
order to obtain data about throughput, queue times, buffer size etc.  These
models are characterised by having numerous variables which can affect the
performance of the system.  Optimising the system requires an understanding of
the interplay of these variables, often gained by adjusting them one by one
over a number of simulation runs and studying the output.  This ad hoc
approach can lead to misunderstanding and sub-optimisation of the system.

Multivariate analysis is a technique which has long been applied to process
control when the interplay between numerous variables (temperatures,
pressures, flow rates etc.) which affect the output of the process is not
obvious.  Multivariate analysis allows the combined effects of process
variables to be understood by identifying those variables which have the most
significant effect on the output and establishing the relationship between
them.

The work presented here describes the application of multivariate analysis in
combination with the discrete-event simulation of a manufacturing system.  The
work showed that multivariate analysis is a powerful tool for obtaining
information about otherwise hidden relationships between system variables. 
Identifying such interplay allows true optimisation of the system to be
carried out.

30. New ways in the plant management: Eberhard Bluemel, Andreas 
Roegnitz, Holger Seidel

Keyword
Plant management, maintenance, simulation, logistics, planning


This paper describes results of several applied research projects of the Fraunhofer Institute of 
Factory Operation an Automation, which were realized in Co-operation with industrial 
enterprises. Effective plant management requires vision beyond the planning up to operating a 
future plant. The simulation as well as tools developed recently to the analysis and evaluation 
allowed the combination to the planning and the operating level and improved the transparency 
in planning and business processes. An essential aspect is the possibility of the examination of 
the consequences of planned measures under operating conditions and the reassignment of the 
knowledge resulting from this into the planning.

The always more progressive concentration production technology with few and always more 
efficient working plants leads to an increasing complexity. The risk is increased through this 
when operating a such plant. For an effective plant management the connection of the 
planning, realization and support of production plants is usefully. This presupposes the 
following for the planning and inauguration of new production plants:

	- Use of available experiences on the part of the maintenance from the operating                                                                                                                    
	  response of plants passing
	- Favorable coordination with regard to the consumption of resources like 
capital, 	  personal capacities and time
	- Creation of cost transparency in the investment and maintenance area
	- Cost saving and performance improvement by the central use of profitable 
	  technologies like critical path method and simulation
	- Creation of a common database for all area bound like also area general 
applications

The decision was due to the extraordinary complexity at the looked processes and the variety 
of stochastic influence sizes met, as planning supporting instruments to develop simulation 
models which give information about the future system behavior and a suitable plant 
management.
 
The aim consisted to develop simulation models and evaluation tools with whose help it is in 
the front-end of the investment decisions and strategy changes possible for an efficient plant 
management:

- to Develop and compare alternatives of logistical imbedding of new plants in 
  the total operating system,
- to Analyse the effects of the behavior of the new plants on the total operating system,
- to Determine reactions to boundary conditions changing themselves (flexibility),
- to Examine alternative measures as reaction to these changed requests  and
- to Work out suggestions for the improvement in the planned techniques and technologies.

The procedure for the development of the simulation models consisted of the following steps:

	Problem analysis/Data acquisition
	Model construction 
	Model implementation
	Validation

For support of modeling the development of tTools followed to the analysis and evaluation of 
the simulation experiments. The aim passed reflow of the won knowledge in the derivation, 
assessment and variation of the substantial influence parameters on the process of the transition 
of the planning to the business of a plant and this one into the planning. 

At the example of a paint shop of the BMW AG became accompanyingly use of the developed 
tools provenly to the three phases planning, realization and business of the plants both in 
production planning and in the working place planning. With the help of the simulation 
experiments of the production system the system behavior was tested under various conditions. 

For the retention of the logistical parameters demanded by the customer (like delivery due 
date, quantity and quality) is the guarantee of the technical availability and the required process 
quality necessary prerequisite. At increased demand to carry out an useful availability increase 
therefore is in a definite period of time.

The result of the availability increase shows the following consequences in the examined 
production system:

	- Influencing of the complete performance of the system regarding the increase 
	  of the throughput
	- Improvement in the process stability by reduction of the fluctuation margin of 
	  the throughput 

Four models made which are different in her maintenance strategies with various technical 
availabilities and in the logistical expiry in a definite partly system became to this. 
The models allow this one the determination at a predefined manufacture quantity to the 
required availabilities. In addition is derivable from the graphic, for which bing aimed crowds 
an availability increase proves as meaningful and represents therefore an effective decision 
instrument for investments or strategy changes. 




31. Configuration synthesis and refinement: an approach to 
simulation of heterogenous distributed systems: Alexander A. 
Lavrov, Leonid S. Yampolsky


Distributed frameworks became an essential paradigm in the design
of modern manufacturing, control, information and other systems.
The real life - oriented development and implementation of such
systems, being a very complex task itself, is often complicated by
their heterogenous nature and intricate synchronisation/coordination
schemes. The available tools for solving synthesis, verification,
and simulation tasks are quite powerful and display numerous useful
features. However, they are usually specialized  (e.g. only
synthesis-, or verification+simulation - oriented), do not support
a throughout (automatic) synthesis-simulation-refinement-verification-
revision line in the scope of a uniform framework, etc.

The proposed approach is based on the interplay between the coordination
structure synthesis stages and the simulation stages. Instead of
implementing a simulation model with subsequent verification of the whole
set of the required/desired properties, now a (possibly maximal) general
coordination structure is synthesized which _guarantees_ the fulfilment of
the _necessary_ set of specifications imposed on the agent synchronisation
structure and represented in terms of a partial order relation and/or
temporal logic. The synthesis proceeds while leaving "irelands of leeway"
for
subsequent refinement and analysis of the system parameters/configuration via
simulation and formal verification techniques. A process-algebraic
description, obtained after the first stage, has a consistent Petri net-based
semantics and thus can be directly translated into the underlying net, which
in turn can serve for both simulation and verification purposes (the second
stage). the results of the second stage can lead to a repeated synthesis
attempt, with revised "limits of freedom", and so forth. The framework which
results from using the approach can be directly translated into a real system
implementation, permits inserting "intelligent" components, etc.



32. Requirements and Concepts for a Production Management Game 
to Be Used in Academic Education of Industrial Engineers: Ruth 
Aguilar, Johannes Krauth, Kim Jansson

Concepts and practices of production management are presently undergoing 
rapid changes. New paradigms and buzzwords come and go so quickly that it 
is often difficult to understand their essence and the difference between 
all of them. Especially students who normally do not have own experience 
find it hard to judge which of these new ideas is useful, and in what 
environment it can be applied.

In order to overcome this lack of experience, we propose to use 
production management games. Representing in a simplified way all 
essential features of an industrial enterprise, games can be a good and 
cheap tool for students to get to know the problems of production 
management, to see their relevance and their interconnection. 

This need for student training can be answered by a simple game 
consisting of production toys organising the shop-floor in different 
ways. The toys can be used to demonstrate different production control 
techniques such as pull, push or bottleneck optimisation. 

Advanced Training

When students or production management people are trained for more 
demanding real word environments, more complicated production systems 
must be used as the bases for a training game. This is not any more 
possible to accomplish by toys, computer game tools are needed.

What are the requirements for such a game? One thing is for sure: In a 
time of rapid change it is not sufficient to have just one well-defined 
game or even a limited set of games. As the management of change is 
becoming an ever more important task of industrial managers, a game must 
also teach students how to manage change. It must provide them with 
methods and tools to evaluate current practise and to improve it, i.e. to 
redesign the business processes and introduce new organisational 
structures.

Location and Time Independent Training

Production management staff are usually very busy people and it is 
difficult to disconnect them from their everyday activities. When several 
persons have to find a common time for a joint training session it 
becomes extremely difficult. Therefore a location and time independent 
training session can be useful.

The Internet provides a excellent media for such a training. Geographical 
distances and different time zones and working hours can be overcome. 
However this media imposes new demands on the architecture of such a 
game. Good help functions and on-line instructions are needed so that the 
games can be used without an teacher. 

Conclusion

All these training types can be combined into a mobile training & 
learning laboratory. Needed accessories, including toys, computer and 
telecommunication equipment,  can be packed in suitcase or equally sized 
package. Our paper explores the requirement for the games in the 
laboratory in more detail, and it derives concepts for production 
management games which can be used on the three different levels of  
education. 

The paper also presents a prototype game which has been developed at 
BIBA. This game presents a number of "standard" organisational concepts 
(tayloristic, group technology, Kanban). Recommendation for further 
implementations of such a mobile strategic learning laboratory will 
conclude our paper.


33. On multistage recourse problems with continuous event 
distribution: Leonidas L. Sakalauskas


    A multistage problem of stochastic linear programming (SLP) with 
continuous event distribution is considered. The expression of a
generalized gradient of the objective function is presented. Then a 
stochastic method for solving SLP, based on this expression, using
Monte-Carlo series, is developed. The rule for Monte-Carlo sample
size is introduced and it is shown that this rule guarantees the 
convergence
with a linear rate. The statistical stopping criterion is also 
considered.
The method proposed is applied to operation planning : in queueing
systems, inventory and location problems. 



34. On the statistical regularity of the indifferent uncertainty : Victor 
Ivanenko, Ivan Zorych

 

  The  problem  of  decision  making  under  the  indifferent 
uncertainty was studied by V.Ivanenko  and  V.Labkovsky.  The 
notion of G[l]- independence was introduced to describe  this 
type of uncertainty. In the subsequent  work  of   the   same 
authors the notion of statistical regularity  was  introduced 
as a mathematical tool  for  the  description  of  a  general 
uncertainty.
  The notion of  G[l]- independence  generalizes  the  notion  
of  probability  independence  and   uses   essentially   the  
properties of the loss-function  l(*,*). It was revealed that 
one had to consider the independence  of  two sequences  only 
with respect to the specific task of decision making.
  The notion of statistical  regularity is  a  generalization 
of the  probability  distribution.  This  notion  allows   us  
to consider the problem of decision  making  from   the   new  
point of  view.   In  this  paper  we  establish  connections  
between the notion of two G[l]-  independent   sequences  and 
the   notion  of  the  statistical   regularity   of    these 
sequences.
  The   example   of   decision   making    problem under the 
indifferent uncertainty is given in the paper.



35. HARD NUT: Educational Simulator of Managerial Functions: E. 
Bagdonas, I. Patasiene, V. Skvernys


Kaunas University of Technology, Lithuania

Object. The simulation software is designed to train students in understanding and 
analyzing real financial statements. Usually the simulation game session covers the period of 5 
fiscal years. The aim of the game is to show the interdependencies between different 
managerial functions, related to various aspects of company activities, such as production, 
personnel, commerce, marketing and finance. The simulator has 19 input variables to control 
the decision making process.

Educational aspect. The simulation game provides the players with a background for 
understanding managerial decisions made under the free market conditions. The software 
allows to tune up the simulator according to the educational background of the players, 
ranging from the high school to university.

Evaluation. The HARD NUT simulator was used in teaching management related 
courses at Kaunas University of Technology and at 8 technical schools in Lithuania. The 
evaluation results for the courses prove that the students’ involvement and participation has 
increased due to the highly interactive nature of the simulation game.

Prospects. The second version of this software based on the more detailed financial and 
market management models is under development.


36. Model development for Energy Tariffs in Latvia: Gaidis Klavs, 
Poul Erik Grohnheit 

Abstract
Within the framework of the energy research programmes of the European Commission a 
series of tools and methodologies has been developed and used for the analysis of strategies in 
the fields of energy planning and environmental protection. Recently the existing network of 
experts and institutions within the European Union has been expanded to include partners in 
Central and Eastern Europe, focusing on transfer of models and training in model usage and 
interpretation of results (the EURIO project). This approach was also used as a part of the 
"Energy tariff project - Latvia", financed by Danish Energy Research Programme, and carried out as 
a collaboration between utilities, research institutes and consultants in Denmark and Latvia.
This paper describes the development and usage of two types of models that were used 
within this project:
•	A system of Customer Calculation Sheets for the analysis of consumer expenditure and 
utility revenue using widely known spreadsheet software. 
•	The EFOM model has been adapted to describe the long term reactions for the electricity 
and district heating sectors in different regions..
These models are used to describe the competition between natural gas, other fuels, and 
imported electricity for the supply of the electricity and district heating sector. They are also 
used to determine the consumers’ expenditures and the revenues for the utilities.



37. The Bootstrap Method and Modelling of Systems: Alexander 
Andronov, Yuri Merkuryev

***Department of Information Treatment and Control Computer SystemsRiga Aviation 
University, 1, Lomonosov St., LV-1019 Riga, LatviaE-mail: rau@rau.vernet.lv**Department 
of Modelling and SimulationRiga Technical University, 1, Kalku St., LV-1658 Riga, LatviaE-
mail: merkur@itl.rtu.lv

A new approach to the modelling of systems is proposed. In the traditional approach, 
observations of input variables are used for estimation of probability distributions of these 
variables. Such estimated distributions are used to generate realizations of input variables 
during modelling runs. Estimation of probability distributions is subjected to making mistakes 
in choosing the form of distribution and estimation of its parameters.The proposed approach 
does not ask for estimation of input probability distributions. Here observations of input 
variables are used themselves, in various combinations, to serve as realizations of input data in 
modelling runs. In the paper, a general scheme (the wave algorithm) of using the Bootstrap 
method in modelling is considered. Values of random variables are not generated during 
modelling, but extracted from input or obtained sample populations. Then the same data of 
populations are used repeatedly in various combinations with data of other populations. It is 
shown that the Bootstrap method makes it possible to avoid the bias of simulation results and 
to decrease their variance. For example, it could be implemented for modelling of system 
reliability.



38. Application of a Queueing Network Model for Analyzing and 
Optimizing an Aircraft Operation: Alexander Andronov, Ivan 
Babichev, Anatolij Kiselenko

The aircraft operation is described by closed queueing networks in the following way. We 
consider an aircraft as a customer of a closed class in the queueing network. There is a fixed 
number of customers. The states of aircraft operation (a flight, a maintenance, a repair, etc.) 
are described as nodes of the queueing network. There is a certain service for each state 
(node). Each service is characterized by a service rate and a servers number (number of 
scheduled flights or maintenance crews, etc.). The operation's states of each aircraft are 
changed accordingly to the irreducible homogeneous Markov chain.The main effictiveness 
criteria of aircraft operation (the average rate of flight, the demurrage of the aircraft, etc.) can 
be computed using the following perfomance measures respective to nodes: the mean queue 
size, the mean sojourn time, the throughput, etc.Suppoze, we can command financial, raw 
material or manpower resources. What distribution of these resouces between operation's 
states ensure the most improvement of the effictiveness criteria? It is a well known problem of 
optimization of closed queueing networks. A mathematical model, an algorithm for the 
computation of the optimal solution and an example are presented.


39 .Training techniques of air traffic controllers' efficincy: Eugene 
Kopitov, Elena Trapeznikova

The goal of the research is classification of professional skills (PS) 
important for air traffic controllers, training and testing technology design.

During the research the four stages of air traffic controllers training 
process were estimated: professional selection, professional training, 
re-training on new facilities, maintaining the necessary PS level. Software 
and methods were analyzed of each stage, with the emphasis of computer 
facilities.

Thanks to the designed methods of testing PS of air traffic controllers, it 
is possible to test anticipation skills, decision making speed, attention, 
memory, sense of time and ability to carry out the psychological condition 
diagnostic. Reliability and validation of designed PS testing were carried 
out. The designed testing was checked and proved effective various groups of 
air traffic controllers.

The research of complex simulator effectiveness for training of air traffic 
controllers was executed. To fulfill general simulation requirements it is 
important to use the science of educational technology when designing the 
educational tools.

The designed technology was approved in the Educational center of "Latvijas 
Gaisa Satiksme".



40. Reliability Dynamic Models Based on Petri Net Application: Igor 
Kabashkin


The acticle presents an approach for technical system reliability analysis 
with Petri Net (PN) modelling. More general and useful method to investigate 
the dynamic behavior of the reliability model under various levels of 
abstraction is discussed on the base of reachability and other analytical 
properties of Petru Nets.

The problems of fault detection and fault propagation are discussed in the 
modelling of fault-free analysis with PN simulation tool. The base set 
including six basic PN modelling elements is described. They allow: 
• to simulate failures with various distributions;
• to simulate alternate consequences of the same failure for various 
conditions of its appearance;
• to simulate mechanisms of process development degradation;
• to simulate elements with two and more types of failures;
• to simulate elements with several levels of reliability.

The present article discusses an application of the basic PN modelling 
elements to represent fault-free. Formalised provedure of Petri Net model 
construction is described. Example of PN model construction is presented for 
self- suficient redundancy repeater station.


41. Aggregative systems: their modelling and simulation: V. V. 
Rykov



A system is an object, which may be decomposed on subsystems
 concerning, at least partially, the properties of the entire system.
 A complex system is that, whose subsystems permit further decomposi-
 tion.  Nondecomposable parts of a system are  referred to as its
 elements. Generally, if the subsystems and elements of the system are
 continuously interacting the representation of the complex system as
 a hierarchical structure is of little help to investigate its
 functioning and control processes.
         A class of aggregative systems (A-systems) which generalize
 the noution of piecewise-linear aggregative systems due to  N.
 Buslenko, V. Kalashnikov and I. Kovalenko [1] is considered.
         A system is called aggregative (A-system) with respect to the
 functions or processes considered if its subsystems of the same level
 exchange information in discrete, generally, random points of time
 only, and operates independently between these points. A complex
 A-system is that, each subsystem of which is itself an A-system.
         For investigation of the functioning and control processes in
 A-systems we use the theory of decomposable and controllable
  semiregenerative processes [2]. The paper contains major definitions
 and results of its theory [3-5] which allow to propose conventional
 tools for simulation of this class of system.
 
                         REFERENCES
 
 1. Buslenko N. P., Kalashnikov V. V., Kovalenko I. N. Lectures on
    the theory of complex systems. Sov. Radio Publ., Moscow, 1973.
 2. Rykov V. V., Yastrebenetsky M. A. On regenerative processes
    with several types of regenrations points. Kibernetika, 1971,
    82--86.
 3. Rykov V. V. Regenerative processes with embedded periods of
    regenerations and their applications in the analysis of priority
    queueing systems. Kibernetika, 1975, 105--111.
 4. Jolkof S., Rykov V. Generalized regenerative processes with embedded
    regeneration periods and their applications. MOS, ser.
    Optimization, 1981, 575--591.
 5. Kitaev M. Yu., Rykov V. V. Conttrolled Queueing Systems.
    SRS-Press. New York, 1996



42. Simulation Assessment of Inter-Organisational Business 
Process Re-engineering Initiatives: George M. GIAGLIS, Georgios I. 
DOUKIDIS, Ray J. PAUL


Simulation modelling and analysis of business processes is increasingly being 
recognised as an extremely useful tool for assessing the feasibility,
efficiency, and  effectiveness of BPR decisions. However, almost all of the
reported experience is  confined to re-engineering initiatives within a single
organisation. The problems that  seem to arise in more complex situations
where re-engineering extends beyond the  boundaries of a single organisation
to include multiple businesses in the value chain,  seem to be quite different
and more difficult to overcome.

In this paper, we argue that simulation can efficiently satisfy the additional
 requirements of inter-organisational business modelling. Based on an
empirical case  study of inter-organisational BPR (involving a multinational
pharmaceutical company  and one medium-sized distributor of its products in
Greece), we identify a set of  requirements for the effective application of
simulation as a decision making tool for  BPR and suggest a framework to
incorporate the distinct characteristics and  requirements for the successful
modelling of inter-organisational business processes.  The simulation model
that was developed for the case study is described and the  usefulness of
object oriented simulation techniques to accommodate the requirements  of
inter-organisational modelling are illustrated. Finally, we derive some
general  conclusions that were obtained from the particular study and indicate
some areas for  further research on the development of simulation models for
business process analysis  in the BPR context.


43. A software for random elements modelling: Victor Galinskij, Yurij 
Kharin

     In simulation of complex processes in economics and
business with the aid of probability models there exists a problem
of random elements modelling (random events, variables, vectors,
processes).
     For modelling of random elements on personal computers, which
are IBM-PC compatible, in Belarusian State University is developed
a package of applied programs (PAP) STATMOD.
     PAP STATMOD can be used for following problems solving:
     - random variables modelling;
     - random vectors modelling;
     - random processes modelling;
     - testing on adequacy of the simulated data to probability models.
     In this report the structure and functional properties of PAP
STATMOD are widely described, and some examples of its usefullness
for solving of applied problems and for students teaching are given.


44. Structuring  and Analyzing Accounting Numbers  Using Self-
Organizing Maps: Barbro Back, Kaisa Sere & Hannu Vanharanta


The amount of financial information in today's sofisticated large databases
is huge and makes comparisons between company performance -  especially
over time -  difficult or at least very time consuming. The aim of this
paper is to investigate whether neural networks in the form of
self-organizing maps can be used to structure and analyze accounting
numbers in large databases over several time periods. By using
self-organizing maps, instead of for example cluster analysis, we overcome
the problems associated with finding the appropriate underlying
distribution and the functional form of the underlying data in the
structuring task. The method chosen also offers a way of visualizing the
results. The database in this study consists of annual reports of 160 world
wide forest companies with data from a five year time period.



45. Reengineering Management: A cognitive approach to 
reengineering: Shigehisa Tsuchiya

Changing the reigning culture, or values and beliefs, in an organization is
the most  important and difficult part of business reengineering.  In
reengineering it is not  sufficient to redesign processes alone.  All four
points on the Hammer and Champy (IU (Js  business system diamond have to fit
together or the company will be flawed and misshapen. 
 The essential purpose, or the result, of business reengineering is vertical
as well as  horizontal compression of jobs or tasks which requires completely
new attitude of  employees.  The rules of governance (and self-governance) for
effective business  enterprises are being determined by their culture.  
	Double-loop learning is necessary to change culture.  It is, however,
extremely  difficult because frameworks for interpreting experience are
generally resistant to  change.  We learn best from direct experience.  In the
real world, we seldom directly  experience the consequences of many of our
most important decisions.  What has happened  is not always obvious, the
causality of events is difficult to untangle, and  interpretations of
experience depend on the frameworks within which events are  comprehended. 
The purpose of this paper is to present a middle range theory on 
effectiveness of computerized simulation for double-loop learning which is
indispensable  for successful reengineering.


46. Simulation of Multi-Agent Real-Time Target Assignment: Harri 
Haanpää, Juha Mäntysaari, Risto Lahdelma

Keywords:  Simulation, target assignment, co-operative agents,
           piecewise linear network problem

We consider the problem of allocating m independent agents to n
distinct targets. Associated to each target is a weight w(j), denoting
the importance of reaching the target. An connectivity matrix a(i,j)
specifies which agents can reach wich target. For each target the
probability p(j) of reaching the target is known. The optimal
assignment of targets to agents is modelled as nonlinear minimum cost
network flow problem, which can be transformed into a piecewise
linear network flow problem.

A preliminary optimal assignment is computed prior to the mission.
During the mission certain significant events may happen, such as
agents becoming disabled, targets being reached, or connections
between agents and targets cut off. In case of such events, the
agents may communicate and re-solve the allocation problem.

We will present a graphical simulator of the multi-agent target
allocation application. The system is implemented in C++ using the
process-oriented SIM2 simulator library, which provides
pseudo-parallel process classes. The network problems are solved
using the NP2 network solver developed at Helsinki University of
Technology. The system runs in PC/Windows systems.


47. A Concept for  Simulation and Optimisation of  Energy Trade: 
Simo Makkonen, Risto Lahdelma

The recent changes in the European electricity market are opening up 
new possibilities for energy companies to improve their efficiency. 
On the other hand the competition between energy companies requires 
that more attention be paid on planning the energy trade. In this 
development, the simulation and optimisation tools and methods have 
come to play an even more important role. The analysis of  trade 
opportunities and the assessing a risk of potential contracts require 
a well-specified concept which supports the decision making in energy 
trading. 

We present a concept, a software tool and a mathematical model of the 
energy system which are used in the short and long-term planning of 
energy trading. The support tool provides functions for the 
simulation and optimisation of the energy supply from multiple multi-
tariff contracts and several production units. The optimisation 
algorithm uses decomposition techniques to solve time dependent 
problems. For demand side management, some specific functions such as 
marginal cost analysis are included to analyse the feasibility of a 
client. In addition, the system features risk analysis for an energy-
supply portfolio based on Monte-Carlo simulation. 

The system can be utilised both for daily operations in energy market 
and for the training of the energy trade concepts.


48. Fuzzy Modelling of Stochastic Flows in ATM Communication 
Systems: Alexandru Murgu, Risto Lahdelma

In this paper  we consider the problem  of modelling the  stochastic flows 
of traffic in ATM networks. The modelling approach is based on fuzzy logic 
and consists of  describing the traffic  occurring in virtual connections. 
The goal of this model is the smoothing  of the statistical  variations of 
traffic patterns around the nominal scheduled values. Some inferences from 
a class of  Gaussian multistep smoothing  procedures are presented in this 
context. Computer simulation results for ATM asymmetrical flows consisting 
of 300 and 1000 bit/packets entities flowing in capacitated links are also 
included.


49. Modeling methodology and tool for business systems: 
Registrator: Baiba Apine, Arnis Kleins, Ojars Krasts, Uldis 
Sukovskis, Artis Teilans, Vita Zviedre

The paper introduces a methodology and a tool for modeling and simulation of different kind 
of large scale business systems with a number of hierarchically structured objects and complex 
information, material or another flows between them. The tool for the discussed methodology 
is named REGISTRATOR  witch is being developed in Riga Institute of Information 
Technology. REGISTRATOR supports the modeling of organizational and/or technical 
systems.Behavior of any active object can be described using a simulation script language 
witch is classified as the discrete event simulation language. The simulation process can be 
animated to get visual impression about system behavior, appearance time of every passive 
object and their routes within the system. During simulation experiments events are traced and 
statistics are collected. The simulation procedure is discussed in details.The main features of 
this tool are:? a relatively small number of concepts used in the building of the static model: 
only five object classes and 16 types of relations between objects, with the terms defined 
intuitive and easily interpreted even by a novice;? REGISTRATOR allows for the 
implementation of different building strategies for the model, including:  classic (top-down) 
strategies,  reverse (bottom-up) strategies,  mixed strategies;? the possibility of checking the 
completeness and consistency of the model and of generating a list of inconsistencies and 
flaws;


50. Evolutions of Methodologies for Information Systems Work: 
Nilsson, Anders G.


In this paper I will illustrate how methodologies for information systems work have evolved in 
the course of time. The concept "systems work" includes both the information systems 
development and software maintenance. My presentation of methodologies has a historical 
perspective and a special focus on Swedish conditions. This review has a strong personal bias, 
but I try to formalize my analysis of methodologies along six dimensions: constituents of 
methodologies, focus on aspects, life cycle coverage, prespecification of systems work, varios 
tools and different views on methodologies.



51. Simulation Supported Industrial Training -Experiences and New 
Possibilities: Jaime Villegas

Abstract proposal for the Riga Conference September 19-21 /1996 


Economics Information Systems
Dept of Computer and Information Science
Linköping University
Jaivi@ida.liu.se

ABSTRACT 

Simulation Supported Industrial Training is a training method which has been extensively used 
in different training projects during last three years. The main idea of this training method is to 
help participants to better understand their own problems at the company with the help of 
computer-based simulation games. 
The method has been created at Linköping University, Economic Information Systems, and it 
has been thoroughly documented in my doctoral thesis ”Simulation Supported Industrial 
Training from an Organisational Perspective -The development and Evaluation of the SSIT 
method”, Dissertation 429, Linköping University. The results have been so encouraged that 
new training projects have already started in companies such as Saab Military Aircraft.

This paper will make a review of the main lessons we have learned in the training projects and 
it will describe the present status of development and use of this method. This description will 
also make emphasis on those new alternatives for the further development of the training 
method. 




52. Developing Simulation Models as Sales Tools - A Case Study: 
Krisjanis Steins

Article discusses the use of computer simulation in business communication,
where simulation models are used as aid in the sales process. Case study
involved development of such models in Swedish company BURMEK, specialized
in manufacturing computerized machine lines for efficient pre engineered
house production. Two simulation models were developed, first by using
simulation package WITNESS, and second by writing custom simulation tool.
Advantages and disadvantages of both approaches are discussed.






53. 




54. Method for Critical States Averting of Monitored Objects: 
Ermizha, Andrejs

Abstract:
Andrejs Ermizha, Dr. comp. sci.

The method under review is devoted to working out the mechanisms of critical 
states averting (CSA) of monitored objects. This method is a part of more 
complex method of interactive analysis of functional models of distributed 
systems which is based on the utilization of up-to-day conception of active 
object-oriented databases for specification, generation and analysis of 
versions of event-driven and time-driven monitoring models (the active 
objects are capable of triggering responding to events when certain 
conditions are satisfied).

The are envisaged the integration of the procedures of object monitoring for 
CSA and their behaviour modeling for data accumulation, actions scheduling 
and performance analysis in the method. The method is destined for 
investigation of model versions (in case of versionable objects) under 
conditions of varying parameters of model and various circumstances of 
enviroment. The first stage of the method elaboration is the level of 
metamodels.

Areas of application of method: investigations of distributed systems with 
processes of signal interactions between objects; in particular, co-design 
of hardware/software distributed systems.


55. Simulation and Modelling at the Riga Harbour Container 
Terminal: L.Novitski, P.Lorentz, E.Bluemel, Y.Merkuryev, Y.Toluev, 
E.Ginters

ABSTRACTS

SIMULATION AND MODELLING AT THE RIGA HARBOUR CONTAINER TERMINAL

Leonid Novitski, Peter Lorentz, Eberhard Bluemel, Yuri Merkuryev, Yuri Toluev, Egils 
Ginters

The paper discusses various aspects  of  using  modelling  and
simulation within designing of Data Processing System (DPS) at
the Riga Harbour Container Terminal  (RHCT).  This  system  is
aimed to  support  terminal operations with modern information
technologies.

Design of the RHCT DPS includes following main stages:
- analysis of data flows and information requirements;
- development of a static model;
- development of a dynamic model;
- design of a database structure;
- analysis and estimation of a functional model;
- models correction.

First, DPS is described as a  set  of  objects  and  important
links. At the next designing level,  a complex of mathematical
methods to represent DPS functioning in dynamics and to create
a data base is used.

For DPS analysis,  estimation of its efficacy and influence to
customers, a simulatiom model is used. The simulation model of
container terminal  technological  processes  allows  to study
different control  strategies  of  terminal  opertation.  This
model is  realized  with the GPSS/H simulation language and is
discussed in  details  in  the  paper.   The  animation   tool
Proof/Animation is  used  for visual representation of results
of the terminal operations simulation.


56. A Simulation Goal Specification Inference Procedure:  Galina 
Merkuryeva

Riga Technical University, LATVIA

Control  of the simulation process is based on schemes of simulation procedure, corresponding 
to the various goals of simulation, e.g. Process Evaluation (G1), What If Analysis (G2), 
Sensivity Analysis (G3), Bottleneck Analysis (G4), Comparison of Alternatives (G5), 
Optimization of System Parameters (G6), Prediction (G7), Metamodelling (G8) or Simulation 
Training (G9). A Goal Specification Inference Procedure for discrete-event simulation studies 
is presented. It is based on specification of each goal incorporating knowledge about the goal 
and the corresponding simulation scheme.
Short 	description of each goal is provided. Tools for knowledge structuring and 
specification of  simulation goals are presented. Examples of specification the following 
simulation goals G1,G4,G5 are described. The corresponding inference procedure is given. Its 
implementation in LATISS Intelligent Simulation System is illustrated.




57. The Role of Manufacturing Planning and Control Systems in 
Process Management: Mikael Ankelius, Jan Olhager and Birger 
Rapp


Traditionally, manufacturing planning and control (MPC) systems have had a resource focus, 
see e.g. MRPII (Manufacturing resource planning) and ERP (Enterprise resource planning). 
Recently, manufacturing re-engineering programs such as time-based management and total 
quality management have brought on a change of focus from resources to business processes. 
Consequently, manufacturing planning and control need to apply a process orientation. The 
purpose of this paper is to identify the business processes that are relevant to manufacturing 
planning and control, and to study the MPC system as a process management support system.


Keywords:
Process management, manufacturing planning and control, business process re-engineering






58. Hydrogeological model for the central region of latvia: Spalvins 
A.

Riga Technical University, Environment Modelling Centre
9 Ausekla Street, Riga, LV-1010, Latvia

The REgional MOdel (REMO) "Large Riga" has been created as a tool for
investigating groundwater reserves and quality of the central part of
Latvia. REMO simulates a multilayered groundwater system including not
only the above region to be closely studied, but the Northern part of
Lithuania and a considerable fragment of the Gulf of Riga too. The REMO
3D-grid contains  N = 41 x 43 x 9 = 15,480 nodes approximating 9 aquifers
separated by semipervious strata. An original Geological Data
Interpolation (GDI) routine has been developed to prepare a REMO
input: a geometry and maps of permeability and boundary condition
surfaces. Algorithms accounting for a groundwater discharge and for
interpolatintg simulation results at observation wells have been
developed. Calibration algorithms for adjusting REMO parameters
have been designed and used.


59. Simulation  of plants for grain primary processing: P.Rivza

Institute of Informatics, Latvia University of Agriculture,
2 Liela street, Jelgava, LV-3001, Latvia, E-mail:rivza@inka.cs.llu.lv

Simulation models may offer an effective remedy while designing plants of grain primary 
processing. 
These models are created for the grain reception section, for the grain preliminary storage 
section, for the dryer as well as for the whole plant.
Two technological schemes were used in the models - the first scheme provides for the 
reception of only one grain flow simultaneously; the second one - for three grain flows 
simultaneously.
The machinery safety is estimated by taking into consideration the elapsed time  between the 
breakdown of the machines and by the time of repair, both of them are simulated with the 
normal probability distribution.
All simulation models, except the numerical outlet date, receive also histograms, e.g., the 
histogram of the number of ventilation bins, the histogram of the relationship between the 
dryer productivity and the grain daily thrashing etc.
The simulation models are most of all susceptible to the changes of the arrive rate and of the 
grain moisture as well as to the machinery safety.
All simulation models are created by making use of the programming and modelling system 
SITA/C, designed by the Institute of Mathematics and Informatics, at Latvia University.


60. Modeling and Performance Evaluation of Production Lines Using 
the Modeling Language MOSEL: G. Bolch, S. Greiner and H. Herold 

Modeling and performance evaluation  of current production systems
gains more and more importance because of growing system complexity
and production costs. Therefore tools and methods are needed that
allow a simple modeling and performance evaluation of such systems.
The goal of this paper is to give a short introduction into the
specification of production systems using the model description
language MOSEL (MOdeling  Specification and  Evaluation Language)
and the analysis tool MOSES (MOdeling Specification and Evaluation
System) which uses MOSEL as input. Therefore the characteristics of
the components of production systems (like batch processing or
unreliable machines) are introduced. For each of these characteristics
a MOSEL specification is given and explained. On the example of a wafer
production system it is shown that using our specification language
also realistic production systems can be specified and analyzed.

Keywords:
Modeling, performance evaluation, model description language,
production lines, Markovian analysis


61. Choosing and optimal data allocation in a distributed database 
system: Djibril Konate

Distributed Database System may be the best choice for computerizing a highly critical to data 
integrity as the State Customs.
The design of a distributed computer system requires taking decisions on the placement of 
data. It's obvious that the placement of data has a great impact on response time, particularly if 
user programs access databases via Wide Area Network (WAN) and the 2PC concurency 
protocol is applied.

There are two alternatives to placing data :
 1. Place data of customs declaration on a central computer and  replicate fully or partialy the 
database to other sites of the network.
2.  There's no central computer. Each site has a portion (fragment) of thedatabase (i.e 
partioned or non-replicated database).

The aim of this paper is to find the optimal data allocation in a distributed system (the State 
Customs) which  I'm designing. In  this paper I develop analytical models of both above 
mentioned methods of data allocation in a distributed database system. The accuracy of these 
models is validated by detailed simulation. The choice of the optimal data allocation is based 
on the analysis of response time.


62. Multivalue Model For Distributed Diagnosis: Al-Jamal Nasser 
(Jordan)

In the capacity of distributed diagnosis model we consider a computer 
network system with interconnected units, which have the ability to test 
each other. In this paper a new model of distributed diagnosis is 
introduced. Suppose that tests are complete, and carried out verdics (by 
test results) are multivalued. Appropriated algorithm of syndrome decoding 
is presented, which allows more exact define technical states of system units.

Multivalue approach presented in this paper may be successfully applied in 
solving of typical computer network problems: routing, performance analysis, 
design and evaluation of topology of communication systems. Depending on test quality and 
testing possibilities, the presented model allows to accomplish complete or partial classification 
of testing system units into classes with different degrees of details.


63. Knowledge representation in the multimodal transport evaluation 
system: D.Dzemydiene, H.Pranevicius, R.Palsaitis

The objective of this study is to analyze the multimodal transport situation between Western 
and Easter Europe and evaluate Lithuania as transit country.

The multimodal transportation may be considered as dynamically
changing and complex technology. Decision making is performed considering a lot of various 
factors: evaluating technical infrastructure, organizational aspects, comparing reports with real 
situation of multimodal transportation.

The results of analysis of the attractiveness of transport systembetween forwading agents and 
carters has showed the most important evaluation criteria: transport cost; reliability; transport 
lead time. The weighting of these three factors varies among the diffrent respondents and was 
also linked to the nature of the cargo being carried.

We have distinguished three levels of knowledge representation in the decision support system: 
the model of informational structure; the model of behavioral analysis of the target system; and 
the model of decision making processes. The deep representation of knowledge provided by 
semantic model allows representation of knowledge provided by semantic model allows.

The deep representation of knowledge provided by a semantic modelallows explicit 
representation of domain specifity and choice of the priorities of principles and rules. The 
conceptual schemes of a topology for road network are suggested. The model of behavioral 
analysis of the system shows the dynamic of observable processes.

The adequate imitation model of the domain behavior allows to prognose further evolution of 
the target system and to increase a quality of decision making.

The task structure relationship with information elements and the course of decision processes 
are represented in the decision making model. A description of this model include the goals and 
represented the strategy of reasoning of specialist-experts in making the decision.


64. Use aggregate approach for business systems formal 
specification and simulation: H.Pranevicius

The theoretical background of aggregate approach is using piece-linear Markof process for 
formal description of aggregates. Aggregate is the mathematical scheme which depends to 
class hybrid automation. The state of aggregate consists from two components: discret and 
continous.

In the intervals when no input signals continous coordinates changeaccording linear law. 
Introduced controlling sequences permit define time instances in which internal events can 
occur. Discret components of state is changed only due input signal or when continous 
coordinate becames zero.

Essential adnevtage of aggregate approach is that it permits to use the same system 
specification both  for creating simulation models and analysis of their correctness. Reacheable 
states and invarinst approaches will be presented for aggregate specification analysis.

The reacheables states analysis approach permits such properties of model as: 1) deadlock 
freeness, 2) completness, 3) termination or cyclic behaviour, 3) boundedness, 4) over 
specification. The invariant approach permits investigate individual properties of model, which 
have to be described  by means of invariant. Invariant is a logical assertion which have to be 
true in each time instant of functioning of system. The approach of developing invariants for 
aggregates models will be presented.

Short information about specification language Estelle/Ag which use aggregate formalism will 
be presented also. How example the aggregate specification of store functioning will be 
analyzed.


65. Use of simulation for planning of work in agriculture: 
H.Pranevicius, A.Kurlavicius

The paper deals with the approach of simulation in agricultural technologic process. The 
developed models permit to make production plans, to analyse production activities, to state 
daily, weekly and monthly assigments. The aggregate approach was used for creating 
simulation models which permits to develop formal specifications of agricultural technologies. 
This method was applied in developing simulation models for poultry and agricultural farms.

The model for a poultry farm enables to plan the flows of birds in structural groups, schedules 
of preventive measures, adjusting them to new production situations, the number of birds in 
age groups, the food consumption, the yield.

The simulation model for a plant-growing farm reflects the tasks, their schedules, the used 
equipment, probability  characteristics of agrotechnical requirements etc.

The results of simulations enable to state the agricurtural equipments to be used, as well as the 
number of machinery and timing. 

By means of the model it is possible to check the stability of the plan, taking into acount the 
changes in  meterological conditions and technical faults.


66. An industrial application of object-oriented modelling and 
simulation: J. Krauth and W. Mühlmann

The paper describes an application of discrete event simulation in an enterprise producing 
noodles. The problem was to design a new distribution and packaging system for a large 
variety of products. Simulation was used to check the general performance of the planned 
transportation system, and to analyse various control strategies. 

The developed solution is a highly customised transportation system with many special 
features. To develop a simulation model of sufficient precision required very specific modules. 
The problem was not only to describe the system behaviour correctly, but also to develop an 
animation which was easys to understand even for persons who had never had any contact with 
simulaiton before ( as was the case with the customers). 

Many of the commercial tools dedicated to manufacturing and transportation systems 
simulation offer very good animation facilities, but do not offer the required specific modules, 
nor do they support the user in developing his own modules. On the other hand, many low-
level simulaiton tools are very flexible with respect to the systemïs behaviour, but have poor 
animation facilities. The applied object-oriented simulation tool Simple++ fulfills both thesse 
requirements: It supports the development of customised modules, and it also offers very 
flexible and easily adaptable animation facilities.

In our paper we will present the problem, the development of the solution, and particularly 
emphasise the role of simulaiton in this process. We will further discuss the advantages and 
problems of object-oriented modelling and the most important features of the applied 
modelling tool.


67. GRAPES-BM Approach to Business Process Reengineering: 
Barzdins, Y. et al.

The paper outlines the business modeling language GRAPES-BM and its support 
tool GRADE-BM.

The language has been designed with two main goals in mind:- formal (for a human) 
description of wide class of business systems, with the main   emphasis on their dynamic 
behaviour and static structure- building executable models of business systems in order to 
perform simulation experiments.

Thus both qualitative and quantitative aspects of business process reengineering are supported 
by GRAPES-BM approach.GRAPES-BM is a graphic language incorporating many features 
of traditional diagramming techniques such as data flow diagrams, event-process chains, etc., 
yet it is distinguished by having precise formal semantics.

The main concepts of GRAPES-BM are:
- tasks representing activities within a system,
events representing anything moving from one task to another (message events) or expressing 
the temporal behaviour of a system (timer events).Large tasks are decomposed into event-
linked chains of smaller ones using Task Communication Diagrams (TCD).

Tasks may have a number of formal properties:
- triggering condition for a task to start
- performer specification referencing elements from the system structure (which is defined   in 
ORG diagram)
- duration
decision branching together with events sent upon the task completion

Stepwise refinement of tasks is supported by many levels of TCD diagrams. Other diagrams 
(Task Specification Diagrams, Datatype Diagrams, ER models, event table, attribute tables) 
help to describe all aspects of a business system.

The support tool GRADE-BM includes- advanced graphical editors supporting automatical 
layout for all kinds of diagrams
- repository management
- syntax analyzer
- simulator coupled with animator and statistics result viewer. They together support:
  * easy visual exploration of business models in action
  * obtaining of default and user defined statistics results of simulation runs.


68. Deterministic Experiments in Optimisation of Stochastic Queuing 
Systems: A Case Study: Vladimir Visipkov 

The paper is devoted to a problem of decreasing the optimisation cost for stochastic simulated 
systems. For this reason, the optimisation process, which is based on a repeated evaluation of 
the goal function,  is split into two stages: deterministic stage and stochastic stage. On the first 
stage, a single deterministic run is used for each optimisation point. In this run all stochastic 
variables are replaced with their mathematical expectations. 

Such an intervention into the simulation process allows us to decrease considerably a number 
of optimisation runs. An approximate solution, resulted from the deterministic stage, serves as 
the initial point for the second (stochastic) one. The discussed approach has been theoretically 
fortified for some classes of the goal functions and has been confirmed by practical examples.

In the paper we generalise the situations in which the application of deterministic experiments 
is reasonable. We also provide an example of the optimisation of a queuing system. The 
analytical solution is compared with the two-stage and one-stage optimisation.




69. Sustainable Development in the Focus of System Dynamics and 
Interactive Learning: Valdis Bisters & Ivars Kudrenickis


Understanding concepts and practise of  sustainable  development requires application of 
variety of tools and techniques. Dynamic link between behaviour of the system and underlying 
structure, social character of decision-making and management of the system is a scope 
covered by the paper. There will be examples of model building and application within 
framework of computer assisted simulation games regarding natural resource management, 
energy planning and economic development.



70. Investigation of distributed computer systems by simulation: A. 
Baums, N.Zaznova

In this report we are representing application of the simulation language SIMAN for 
investigation and comparision of different distributed systems. A library of simulation programs 
for the most typical real-time computer systems (autonomous and distributed) has been 
developed. The special emphasis was made on investigation of local-network different access 
methods and on ways of their improvement. Investigation related to formalization of the 
transition process from the computer system to its model has been carried out. The reported 
research is a part of a project in automatic design of real-time distributed systems.



71. Analytical-Statistical Estimators for Reliability Analysis of 
Cumulative Systems with Time Redundancy: V.D.Shpak

One of the effective ways to overcome various analytical difficulties is to 
use analytical-statistical estimators. A general approach to constructing 
such estimators for transient reliability
indeces of semi-Markov systems of the special case is the proposed in this 
paper.

When analyzing the system reliability only two alternating states of system 
are often recognized up(u) and down(d). At the initial point the system is 
supposed to be in state u with probability one.

The state sojourn times of system are usually assumed to be identically 
distributed random variables with distribution function  F(t) for state u 
and G(t) for state d. In this case the system's behaviors is described by 
some alternating renewal process (let us denote it by v(t)) with states u 
and d (v(0) = u).

An analytical-statistical approach to solve this problem is proposed in the 
paper.

72. Weakly-Formalized Problems. The system "Assistent of Expert" 
and its use in some natural-scientific applications: P. Makagonov, M. 
Alexandrov, V. Denisov & G. Lozovoi

Authors : P. Makagonov, M. Alexandrov, V. Denisov, G. Lozovoi
 
 A b s t r a c t
 
 The class of weakly-formilized problems
 ---------------------------------------
 This is the problems of formation of notions  and  concepts,  connected
 with study of new objects. Source data are matrixes: objects-attributes.
 The  approach  to  decision  is  data  STRUCTURIZATION.  Really,   when
 expert  has  got  object  structures   he   can   put   forward     the
 hypothesises  about  concepts,  which has  united the objects  in  this
 structures, as well as he can to form notions for description  of these
 concepts. This gives him  the  axiomatic  basis for future conclusions. 
 The requirement : data structurization should not be based  on  already
 existing representations ! Therefore methodology  of  decision  is  the
 next :  non-formal procedures of classification  and formal algorithms,
 oriented on data.
  
 'Assistant of Expert'
 --------------------
 This system is SPECIALLY ORIENTED on generating of notions and concepts.
 The system  allows  to  process  the  matrixes, containing  hundreds of
 objects   and   attributes. The  attributes  can  be  quantitative  and
 QUALITATIVE  ones. The  key  position is  graphic dialogue with expert. 
 It allows :
 - to search and to select the clusters in subspaces of factor  space
 - to select the clusters, changing  the distance scale on dendrite 
 - to check  up  the stability of clusters, using both methods
 Clusters  can be  linear,  ring  and  etc.  Clusters  can be located on
 several branches.The skill of expert - to find out them. Classification 
 can be carried out  also  for  attributes because of the convertibility 
 of objects and attributes in matrix. While classifying  it is  possible
 to use  the  geographical  coordinates  of  objects  on the maps. These
 opportunities  give  the  additional  material  for  hypothesises. 
 
 Applications
 ------------
 Two examples from geology are considered: the search of gold and search
 of petroleum.   Two examples from biology are considered:   demographic
 genetics and hereditary illnesses diagnostics.  All examples  use   the
 real experimental data.

73. Modelling without model: Object-Oriented approach and analysis 
of the results by the system "Assistent of Expert" (On example of 
some problem of population genetics): M. Alexandrov, V. Denisov, P. 
Makagonov & G. Lozovoi

Authors :  P. Makagonov, M. Alexandrov, V. Denisov, G. Lozovoi
 
 A b s t r a c t 
 
 The  title  of  paper is not exact intentionally :  the  authors  wanted
 to  attract the  attention to stage 'premodeling', when it is  necessary
 to put forward hypothesises about models of new processes  or  phenomena 
 at  small quantities of initial data.
 
 The construction of imitative model
 -----------------------------------
 The lack of initial data  can  be  compensated  by  imitative model. The
 object-oriented approach  provides  the representation of this model  in
 form of  set  of  objects,  relating to different classes of abstraction.
 As far as  the internal structure of objects is unknown, each  class  is 
 described  by  external observable attributes,  frequently  belonging to  
 various  fields  of  knowledje  (social, biology, technical etc.).   The
 result of simulation is the matrix: objects-attributes. We  often   have
 a hundreds of objects and their attributes.
   
 Ability of 'Assistant of Expert'  
 -------------------------------
 The system is qualifier, enabling expert  to  allocate manually   steady
 groups of objects ( clusters ). It  isn't  known beforehand the quantity
 and images of clusters. The   essential   features of  our procedures of
 clusters-analysis are the  following:
 - Classification is carried out  in  no  attributes space  but in factor
 space,  where  the  expert  chooses  convenient   subspace  for  himself
 - Classification is carried out on dendrite, where expert  cnanges  root
 of tree  and scale of distances.
 The main idea is to  find  out  the  latent  relation  between   objects
 (attributes). On the basis of these  relations  the  hypothesises  about
 internal links in model of objects are put forward  and checked up.  
 
 The example of model restoration
 --------------------------------
 The model of some population development is considered. Model  reflects
 some regularities, connected with diseases of individs, their  behaviour
 etc.  It  is  shown  the  process  of   simulation  and the abilities of 
 'Assistant of Expert'.



74. The Estimation of City Management Quality: Analysis of letters to 
Moscow Mayor's office by the system "Assistant of Expert": P. 
Makagonov

Author :  P. Makagonov
 
 A b s t r a c t 
 
 The   letters   of   citizens  to   the  mayor   with  complaints  and
 proposals on improvement of urban administration work can be  used  as
 an  important  element  of  a  feedback  in  a control system of urban
 services for drawing up operative  administrative  decisions  and  for
 perfection of city normative base.
 
 The binding of a type of the  complaint to an address of the letter's
 author permits to separate problems,  requiring decisions on urban or 
 even at the federal level, from urban region and microregion problems.
 
 Data   of   complaints   in  valuation of activity  of  various  level
 urban  administration  must  be  analyzed  with  taking into account a
 discrepancy of opinions of  correspondents  and  urban  administration
 about  the  level  of management,  which has tools for the decision of
 problems,  mentioned in complaints.  As well it  must  be  noted  that
 distribution  of  items of complaints on urban regions depend not only
 on quality of administration work,  but on  historical  and  long-term
 factors of urban region development also.
 
 The   quantitative   distribution  of  complaints  types   corresponds
 in  rough  approximation  to  the  Zipf law.  An opportunity exists to
 improve  heading  structure  with  the  help  of  the  program,  which
 allocates in  weak-structured texts similar information about a circle
 of problems in a city,  as well as  to  receive  a  picture  of  links
 narrowness   of   urban  territory  sites  and  their  problems  types
 -headings.
 
 The  technique   of  the  analysis of  complaints  is  based  first of
 all on use of a dialogue system,  constructed on algorithms of cluster
 analysis  for  matrix  of  objects and attributes (properties),  where
 territorial units of a city act as objects,  and types  of  complaints
 act as properties.


75. Hyperknowledge as a paradigm for active DSS: Christer Carlsson


IAMSR / Åbo Akademi University
DataCity A 3210, 20520 Åbo, Finland
christer.carlsson@abo.fi


Abstract

We have shown that by combining fuzzy logic and hyperknowledge we will get a new 
paradigm for the development and use of active DSS, which is a more effective and productive 
way to build decision support systems. We demonstrate that this will help us to gain a better 
understanding of decision making, problem solving and planning processes in environments, 
which have not been accessible for systematic studies with traditional research instruments. We 
also show that active DSS tools can be used as research instruments.

A hyperknowledge platform should allow for quick and flexible prototyping, which makes it 
possible for the users to immediately test their intuitive understanding of the systems 
constructs. This can be enhanced to an environment for interactive groupwork on problem 
solving, planning and decision making.

With the help of visual tools the DSS can be built in a way that significantly enhances the 
productivity of the user: proper modelling tools, effective methods for analysis and synthesis, 
and user-supportive tools for communication, data access and data handling, simulation, 
graphics and professional reporting.

When working with a number of management teams (more than 60 managers in about 20 
strategic business units in 4 major corporations) we learned that an active DSS actually could 
be used as a research instrument
 

Fig. 1 Active DSS as a research instrument.

In this set-up the active DSS [A-DSS] is allowed to interact with all parts of the research 
process: the manager, the research team, the theoretical framework and the business context. 
As we have hyperknowledge elements included in the A-DSS there will be traces of all 
interactions which can be analysed and explained after the interactions have taken place. We 
can, for instance, have (i) explanations of conclusions the manager has made with the help of 
the A-DSS, (ii) traces of conclusions about the business context going back to the theoretical 
framework, (iii) comparisons of conclusions the research team has made with the conclusions 
the manager made with the same data, (iv) joint explorations of possible cause-effect 
relationships identified from business data, etc. If we include a log function in the A-DSS, as 
we did in the Woodstrat, we can easily collect data on all these activities - also over extended 
periods of time - and carry out longitudinal studies.

Summarising all these insights the following outline of an emergent paradigm for active DSS 
becomes visible:

i.   decision support should be meaningful, relevant and productive for the intended A-DSS 
users; these users could be both senior managers and managers from all levels of  organisation; 
ii.   the A-DSS technology should allow users to share insights and combine problem solving 
skills, both same place-same time, different place-same time and different place-different time;
iii.  the A-DSS should offer a platform for interactive groupwork on problem solving, planning 
and decision making;
iv.   the software platform for A-DSS should be chosen and/or developed to offer productive 
and effective support for the actual users: proper modelling tools, effective methods for 
analysis and synthesis, and user-supportive tools for communication, data access and data 
handling, simulation, graphics and professional reporting.
v.   the A-DSS should be used as a research instrument and be used to validate and verify the 
theory basis of the management activities being supported.

References

[1]  Carlsson, Christer  Knowledge Formation in Strategic Management, Proceedings of the 
HICSS-27 Conference, 1994, 221-240
[2]  Carlsson, Christer and Walden, Pirkko  Strategic Management with a Hyperknowledge 
Support System, Proceedings of the HICSS-27 Conference, 1994, 241-250
[3]  Carlsson, Christer and Walden, Pirkko  Cognitive Maps and a Hyperknowledge Support 
System in Strategic Management, Group Decision and Negotiation, (1995, forthcoming)
[4]  Carlsson, Christer and Walden, Pirkko  On Fuzzy Hyperknowledge Support Systems, 
NGIT’95 Proceedings, Tel Aviv 1995
[5]  Chang, Ai-Mei, Holsapple, Clyde W. and Whinston, Andrew B., Model Management 
Issues and Directions, Decision Support Systems, Vol 9 (1993), 19-37
[6]  Zadeh, Lotfi A.  The Role of Fuzzy Logic in the Management of Uncertainty in Expert 
Systems, Fuzzy Sets and Systems, Vol 11 (1983), 199-227
[7] Zadeh, Lotfi A.  Knowledge Representation in Fuzzy Logic; in: Yager and Zadeh (eds.): 
An Introduction to Fuzzy Logic Applications in Intelligent Systems, Kluwer Academic Publ, 
Boston (1992)

76. Computer aided business games as a tool for training 
entrepreneurs. Experience of Latvian business consultants: Janis 
Baltacs


Experience of Latvian UAC (Enterprise Support Centers) in designing,
developing and providing training in management, finance management and
marketing for company managers in Latvia.
Who are participants? Why do they attend training? When do they do that?
How and in what way do they perceive and evaluate training? What should the
product look like?





9