MATERIALS

SCIENCE &
ENGINEERING
A
ELSEVIER Materials Science and Engineering A196 (1995) 109-120

The use of computer simulations in training

Peter van Rosmalen, Jan Hensgens

Research Institute for Knowledge Systems Ltd (RIKS), P.O. Box 463, 6200 AL Maastricht, The Netherlands

Abstract

The premise underlying the SAM project is the importance of simulations. Simulations are used to build up understanding by ex-
perimenting with different scenarios and to exercise tasks in an efficient and safe way before doing the real job. State-of-the-art multi-
media tools allow a realistic and appealing front-end and give the possibility of representing information in the most appropriate way.
In this paper we present the SAM environment in which a set of existing and SAM developed tools are offered to create and use ad-
vanced simulation-based learning environments. The aim of the SAM project is to specify and develop a modelling, authoring and
learning framework based on state-of-the-art software tools. The framework will be completed with specific tools supporting the
learner, the author and the modeller. The backbone of SAM is its architecture. It makes it possible to use several cooperative applica-
tions, all operating in the same windowing environment. Each application provides typically only a few of the functionalities. In addi-
tion inter-application communication between the different applications is used. It allows the use of existing applications, e.g. existing
models or multimedia toolkits and can ease the production and learning process by relying to a large extent on known applications. The
SAM Development Environment contains three levels. At the first level, the global design of both course and model is given containing
the course structure, the instructional strategies and the purpose of the model inside the course. The second level contains the specifica-
tion of Instructional Objects. The third level contains all simulation and multimedia material. The SAM Learner Environment distin-
guishes five categories in which learners may require additional support while studying. These categories are: controlling the course
and information space, understanding the subject and retaining that knowledge, analysing and displaying of simulation data, explora-
tory learning and communication. For each category we describe one or more tools that can support the learner.

Keywords: Simulation; Computer assisted instruction; Intelligent tutoring systems; Multimedia; Authoring; Modelling; Learner tools;
Monitoring; Software control; Communication architecture

1. I n t r o d u c t i o n mises the learner control. The degree o f available learner

control defines the perceived level o f interactivity o f the
1.1. Background course [3]. Learner control is seen as the control over
learning strategy, manipulation of learning content and
The importance o f simulations is widely appreciated. description o f content [4]. The claims are that an active
They are used to support decisions and understanding by learning approach facilitates learning. Learners appear to
giving the possibility o f experimenting with different sce- be more engaged and to have better motivations. Simula-
narios. The current d e v e l o p m e n t o f multimedia allows a tions enable learning by doing, knowledge and skills can
realistic and appealing front-end. It gives the possibility be acquired in an active way. Learners can make their
of representing information in the most appropriate way. own errors and learn from these. They can explore the
The rationale behind the use o f multimedia in educa- modelled domain. They can do some experimenting to
tion is that some m e d i a transmit certain kinds of informa- generate and evaluate hypotheses, etc. Moreover, simula-
tion better than others [ 1]. This makes it possible to give tions are powerful in situations in which it otherwise
media an extremely important role in the context of edu- would be difficult (Table 1) or impossible to give training
cation and learning [2]. The motivation for the use of or education.
h y p e r - m e d i a and simulations in an educational context is Experiences in the instructional practice have shown
that it supports an active learning approach and maxi- that these a priori claims on learning facilitation are incor-

0921-5093/95/$09.50 © 1995 Elsevier Science S.A. All rights reserved

SSDI 0921-5093(95)09917-4
110 P. van Rosmalen, J. Hensgens / Materials Science and Engineering A199 (1995) 109-120

Table 1 can be done automatically or manually: automatically by

Examples of situations in which simulations are highlysuitable having an algorithm that selects the learning material that
fits best the current learner's knowledge and profile;
Instead of Real Practice manually by giving the author a way to activate or de-
Dangerous situations Problem scenarios, e.g. a melt-downof a activate parts of a course depending on the expected
nuclear power station
Availability Astronaut training on devices learner's knowledge and profile.
Expensive Equipment, material Within the SAM project we want to achieve a frame-
work which integrates simulations in a multimedia learner
Extend the Real Practice environment with some additional functionality. At the
Reduce or enlarge time Chemical reactions moment there is a wide range of products available cover-
Reduce complexity Economy
Visualise Chemical structures ing parts of the requested functionality. These are dedi-
Redo Building electronic circuits cated simulation environments, multimedia editors, hy-
pertext and computer-based training tools. In our opinion
it makes no sense to re-implement this functionality. A
technical solution has been found to integrate these exist-
rect. The best learners take most advantage of an active ing packages in one SAM environment. However, the
learning approach. This is not surprising because the top final environment must be more than the sum of these
20% of learners will learn whatever we do to them [5]. packages. The functionality not covered must be imple-
For the other learners, encountered problems can impede mented and solutions must be found for known problems
the presumed positive learning effects. Some known belonging to the integrated packages concerning naviga-
problems of hyper-media are on navigation, planning, tion, structuring, orientation, etc.
annotation and the lack of orientation. When the learner is
"lost in hyperspace" the questions that arise are: Where 1.2. O v e r v i e w
am I? How do I get to any destination? and What can be
seen from where I am? [6]. The same feeling of being lost In Section 2, we discuss the needs of the author and
can occur when learners are just put behind a simulation modeller and describe the three levels of the SAM Devel-
without additional assistance. Questions that will arise opment Environment. In connection with this we discuss
are: "What should I do now?", "What does this result the tools we supply to the author and modeller. In Section
mean?". It is realised that simulations need to be embed- 3 the SAM Learner Environment is introduced. The cate-
ded in an instructional environment to fulfil an instructive gories in which learners may require additional support
role in a satisfactory manner [7]. This embedding can be while studying are explained and for each category we
done by putting instructional support around the "un- describe one or more tools that can support the learner.
touched" simulation (like a sandwich) or by interweaving In Section 4 we explain how a development team can
the instructional support with the simulation itself [8]. instantiate their personal SAM environment. SAM is not a
The first option is only feasible when the learners can be fixed environment. It depends on the preferences of the
expected to have enough domain knowledge to manage development team. It is possible to compose a project-
the simulation on their own. In most cases the second dependent development environment. In order to explain
option must be chosen. This means that the simulation how this is achieved, we introduce the SAM technical
must be adapted by order of the author, who designs the architecture and we give two examples of instances of the
actual support. However, adaptation of the simulation SAM environment.
also requests an active involvement of a modeller, who We end our explanation with an example of a small
knows the insides of the specific model. demo course built in SAM. Finally, in Section 6, we give
Users of computer systems appear to be heterogene- an overview of the evaluation results so far and our future
ous. They differ in important features like background plans.
knowledge, levels of intelligence, cognitive style, endur-
ing influences, such as previous education and culture, 2. Developing a simulation-based learning
personality factors, capability, learning ability etc. [9]. environment
We can expect that this heterogeneity will also hold for
courseware, whenever it is developed for a larger audi- 2.1. Introduction
ence. This implies, however, that the efficiency will be
limited as long as its content is defined in a static way. Within the SAM project we want to achieve a frame-
The introduction of dynamic and adaptive courses will work which integrates simulations in a multimedia learner
improve this situation drastically. In our opinion the most environment. At the moment there is a wide range of
suitable adaptation for courseware will be done based on products available covering parts of the requested func-
the learner's knowledge and his/her learner profile de- tionality. These are dedicated simulation environments,
termined by the chosen features. The actual adaptation multimedia editors, hypertext and computer-based train-
 P. van Rosmalen, J. Hensgens I Materials Science and Engineering A199 (1995) 109-120 111

ing tools. In our opinion it makes no sense to re-imple- will be done in the applications that authors and/or mod-
ment this functionality. Therefore we started with a tech- ellers find most appropriate.
nical solution to integrate these existing packages in one Before we deal with these different levels in the fol-
environment (cf. Section 4). However, the final environ- lowing sections, we remark briefly on the structuring of
ment must be more than the sum of these packages. the design. During the design process, emphasis and sup-
Functionality must be implemented and solutions must be port is given to the top-down approach. The specification
found for problems not solved in existing packages, e.g. of course structure, the choice of a model, the specifica-
problems concerning navigation, structuring, orientation, tion of the use of the model and the implementation of the
etc. learning material is done in this order. However, a limited
The SAM implemented parts of the system will lead to bottom-up approach starting from existing learning mate-
the following results: rial is also possible. These limits can be reduced in future
- SAM methods and techniques are specified in rules when the databases with simulations and instructional
and tools to enable the technical and conceptual inte- objects are evolving and clear mechanisms for storage
gration of simulations in courseware. and retrieval are introduced. The global design is de-
- The structure of the course can be made explicit and scribed in two different sections: Section 2.2.1 stresses
surveyable to support the author with a structured the design of the course structure and Section 2.2.2 the
course design and to prevent learners from getting design of the simulation and its use. Section 2.2.3 de-
lost in the course space. This surveyable course scribes the creation of SAM Instructional Objects. The
structure provides the learner with feedback about specification of the monitoring process is discussed in
his/her browsing path and can be used as an always Section 2.2.4 and the creation of multimedia material in
accessible recovery point [10]. The author now can Section 2.2.5.
design an adaptable course in one structure and
compile it into an individualised course to meet the 2.2.1. Global design: course structure
heterogeneous learner. In practice authors often specify the course structure
- Active support for the learner can be provided within on paper by making outlines and diagrams or by drawing
SAM. This includes monitoring of learner actions pictures. With this specification, the course will be im-
and different states of the simulation. Monitoring plemented and after completion, the control flow consti-
data will be used to give feedback and to guide the tutes the (implicit) course structure. From that time on-
learner through simulation and/or course space. The wards an easy inspection of the structure appears to be
author has to specify this monitoring process and to problematic or even impossible within the current gen-
link it with the actual learning material. eration of authoring packages. Moreover, adapting a
- Existing pieces of learning material can be re-used to course to specific learners which could be required to
assemble new courses just by putting them into other serve heterogeneous groups of learners, asks for tedious
course structures. re-programming in the authoring package.
In the following sections we concentrate on the devel- In SAM, the course structure will be specified explic-
opment perspective. In the next section, a map is made itly in a course plan which is formed by a hierarchy of
between the different needs of authors and modellers into Learning Goals, Instructional Plans and Instructional Ob-
tools. jects (Fig. 1). A course will be decomposed in Learning
Goals. Each Learning Goal is linked to one or more In-
2.2. Needs and tools structional Plans describing different ways to achieve the
same Learning Goal. These Instructional Plans in their
The design and development of courses will be done at turn are decomposed into Instructional Objects and/or
three levels. At the global design level, course and model Sub-Learning Goals. A course specification is supposed
are defined as well as how the model(s) can be used. The to be ready when all Learning Goals can be achieved by a
global design contains course structure, instructional set of Instructional Objects.
strategies and a conceptual integration of the chosen Learning Goals, Instructional Plans and Instructional
model(s). At the next level we specify activities or so- Objects can be specialised by attributes. There are two
called Instructional Objects which can be executed in the types of attributes. The first type can be used to put
SAM environment. One of the objectives of SAM is to constrains on all possible orders, e.g. learning goal 1 must
enhance simulation-based learning systems with adaptive be achieved before learning goal 2. The second type
learning facilities. To enable this, the existence of an ac- describes the learners, e.g. their prerequisite knowledge
tive monitor module is a prerequisite. What may be or preferred learning styles. The authors themselves
monitored will be specified by authors. They decide also define these attributes as well as valid sets for their val-
how its outcome will be used to supply, e.g. additional ues.
feedback to the learner. At the third level the simulation Some major advantages generated by our approach are
and all necessary multimedia material is developed. This better support and a structuring of the design task, the
112 P. van Rosmalen, J. Hensgens/ Materials Science and EngineeringA199 (1995) 109-120

Col'~l:
A~lpm,+~e ~oJJ~r ,. e+~ l . i c
~=IZ I n s t r u c t l o n a l Plan [ d l t o r

ij+
U~l.ers~,J w o F k ~ l r ILs~P,,r [] Understand worklng of boiler
of)oi]*r I
0 P l a n I: T e a c h w o r k i n g o f
+ , [] Heater
'1"*J~ 1~workkqT o~" I [] Thermostat
i P l a n 2: ....
ZlSrul<.,--lO]ilwt: 111slr u t : I n x l 0 ]kJ- - t :
iI i| Search for:
TX+rmc~t& ~Lkr 'i[
+tic kalt Jl L J

Fig. 1. A course on boiler mechanicsdecomposedin Learning Goals, Instructional Plans and Instructional Objects. To the left a representation made
on paper. To the right the same decompositioncreated with the help of the InstructionalPlan editor.

storage of design and documentation in the same envi- 2.2.2. Global design: specifying and describing
ronment as the final course and the possibility to reuse simulations
parts of a course on design levels. In addition, the intro- Simulations are useful whenever tasks to be taught are
duction of different Instructional Plans in combination complex and/or learning processes need a context of real
with a possible specialisation by attributes of Learning world settings. Such setting can be approximated by a
Goals, Instructional Plans and/or Instructional Objects well designed model. Including a simulation in a course
enable automatic compiling of several individualised provides the learner with an exploratory environment,
courses out of one course structure. In this way authors which stimulates "the application of certain learning
specify alternative paths through the course. processes, like problem solving, discovery learning and
SAM's use of an Instructional Plan is based on previ- inductive learning" [14]. Whenever authors believe
ous research. An instructional planner [11,12] optimises simulations would fit in the course, the existence of suit-
the learning process by figuring out what, when and how able models or the feasibility to build them must be in-
to teach. A study advisor [13] does a similar job, however vestigated. This investigation requires at least a draft
it will not take final decisions for the learner about what, specification or global design of the model. Two different
how and when. It merely gives users advice on all current specialists take their role in this specification process: the
opportune possibilities. The users themselves may then author and the modeller. The author, responsible for the
decide. Both approaches depend on the same information overall design and development of the course, takes care
(i.e. information on the learning goals to be achieved and of the listings of domain concepts. The modeller is re-
the state of the learner model) and have an equally com- sponsible for a final and adequate technical specification
plex task (i.e. deciding on a large number of knowledge and searches for a suitable model or for the development
sources, part of which changes dynamically). of a new running model. This implies that the global de-
The Instructional Plan module in SAM behaves like a sign of the model must be a collaborative job of author
(simple) study advisor. A specialised tool is designed and and modeller. The common specification is not a trivial
implemented to provide the required functionality. At task given the often different viewpoints and approaches
authoring time, this tool will be used by the authors as an of author and modeller.
Instructional Plan Editor (IP-editor) to construct course The author is mainly interested in what can be learned
structures and to link Instructional Objects to this struc- from the domain and its concepts. The usefulness of a
ture. In addition the IP-editor allows the author to specify model is determined by its ability to clarify important
learner profiles before a course is generated. The IP- domain concepts, showing their changes in time in rela-
editor enables the creation of adaptive courses in a cost- tion with other concepts. These changes must be visual-
effective way. At execution time, learners will have per- ised during a simulation run. Modellers are more inter-
sonalised courses. They gain, depending on their individ- ested in an executable model, which is a reliable domain
ual profiles, access to a subset of the learning material representation. During specification phases bridges must
with a pre-set path. However, learners are not forced to be built between these different viewpoints and the cho-
follow this path. They can control their own path and sen point of departure is the viewpoint of the author.
progress with the running counterpart of the IP-editor The exchange of information about the model can be
namely the IP-browser (see Section 3). done by a direct meeting, but also, as we propose, by the
 P. van Rosmalen, J. Hensgens I Materials Science and Engineering A199 (1995) 109-120 113

creation of Concept Networks [15]. Representations ferent packages. Branching on the returned results is
similar to these concept networks have been implemented made possible. Introduction or feedback material consti-
or are suggested for authors and modellers to support tuted by other instructional objects or ISOs can also be
their work (e.g. the PSAUME interface in [16]). The CN- integrated into an ISO. A graphical flow-chart editor is
editor enables an interactive drawing of hierarchical con- developed to support the author with the specification of
cept networks, which create the specification and the use ISOs control flow (cf. Fig. 10 for a screen snap of the
of a model in a course. Annotations can be put on all ob- ISO-editor and an example ISO). This ISO-editor enables
jects constituting these networks and this enables the the selection of externally created instructional objects
structuring of necessary author-modeller communication. and different modules (simulation, learner tools, monitor
As a next step a CN-editor can support the integration of etc.) with their functions and data to be interchanged, etc.
the model in the course by enabling the needed mapping and to be positioned in the flow chart.
between domain concepts and model entities. ISOs containing a simulation are subdivided into dif-
The use and reuse of externally built simulations is a ferent groups depending on their different levels of free-
tedious task. First, the simulation package must be inte- dom, ranging from fully system guided task exercises
grated into the SAM environment, which is only possible to learner controlled experiments. The groups are: demos,
if the package supports external data communication and tasks, hypotheses, experiments and explorations (cf. [17]
accepts external control (see Section 4 for a discussion on for a more detailed description of the different groups).
how to integrate a simulation package) Second, a SAM- Each simulation ISO will have a connection with a
specific model description must be specified to ensure specific (subview of a) model. At run-time the model will
access to all model variables. be executed and controlled by the ISO. During the design
The author starts with the specification of the needed of the initial state settings, available learner actions and
simulation. The modeller will also, as said before, be en- all monitored conditions must be specified. The ISO-
gaged in the specification process. Whenever specifica- editor provides the needed functionality and enables
tions are finished the modeller searches for an existing authors to design, edit and test the ISOs. ISOs can with
simulation. Found simulations will be adapted or new their scripts, be linked to expected learner behaviour
ones will be built according to the specifications. The (tasks). These tasks will be used by a monitor to recognise
resulting models are made SAM-compatible and are learner actions and/or simulation state changes. The next
stored with their descriptions. These descriptions contain section discusses the specifications of the monitoring
package and model specific information to enable reuse in process.
the future and are made visible during the selection proc-
ess. The information describes the use of the model, its 2.2.4. Specification o f the monitoring process
exported variables with their data types and all available Monitoring in the context of simulation based learning
commands. environments means recognising patterns and sequences
of learner actions and/or state changes of courseware
2.2.3. Creating Instructional Objects and/or simulation. The monitoring can be divided into (a)
Instructional Objects are the basic educational units time based monitoring of variables, (b) checking for vio-
used to develop courses. Each object is a specification of lations of specific conditions and (c) recognising expected
a complete piece of learning material, which can be exe- learner behaviour. In most situations, combinations of
cuted during the computer supported instruction. These these different manifestations will be used. Authors must
Instructional Objects serve several functions, e.g. expla- be able to instruct the monitor what must be monitored
nation, instruction, test and experiment. Their content and what kind of results are needed. The results are used
carders can vary, e.g. text, picture, movie, simulation or in the control flow of an ISO to determine feedback or to
any combination. In a multi-application platform, as is the adapt the instruction. Monitoring can result in recognised
case in the SAM project, the Instructional Objects are plans, in announcements concerning most likely plans,
created with the help of various applications, e.g. Tool- whenever progress is made, or when (possible) plans are
book, MsWord, HyperCard, Authorware. To enable an changed, deviated from or have failed. For the interpreta-
easy sequencing and to guarantee smooth transitions be- tion of the learner actions and state-changes we will use
tween different applications Instructional Objects are in- an approach, called the task-plan approach, similar to that
tegrated in higher levelled objects containing scripts. used for projects like MATIC [18] and ITSIE [19]. The
These objects are called Instructional Scripted Objects approach uses an explicit representation (expected behav-
(ISOs) and constitute the SAM-compatible building iour) of learner actions and state-changes. The Monitor
blocks. classifies actual learner behaviour by comparing it with
The actual integration of a simulation to be used as a this expected behaviour as represented in a task-plan hier-
reusable SAM building block is done within an ISO. This archy. Specialised tools (Fig. 2) have been created for the
ISO starts the simulation and waits until the control re- author to provide the monitor with the expected behav-
turns, which implicates that the control flow crosses dif- iour input.
114 P. van Rosmalen, J. Hensgens / Materials Science and Engineering A199 (1995) 109-120

iHHDE-OH MRIM.VI LabVIEW.c&e

itrue
i

ifalse
t=O : t=
point: ( 3.97~ true )

Fig. 2. Two primitives are defined. A learning action (the two rectangles): "Put switch HHDE on between time <3.9-4.2>". A simulation state (the
parallelogram): "The Temperature has to remain between the specified range".

The task-plan approach has its roots in the Command 3. The learner environment
Language Grammar ( C L G ) of Moran [20]. However, the
main difference with CLG is that the physical compo- In SAM we analysed and documented the learner's
nents of the device and/or program are neglected. Learner need during the exploration and the use of a simulation-
behaviour is represented as a combination of tasks repre- based learning environment. Five categories are distin-
senting possible goals learners might pose themselves guished in which learners may require additional support
in a given context and plans representing the way (authors while studying. Depending on the learning and learner
think that) learners think the task can be accomplished. A involved the author has to decide in which category
plan can be correct and efficient but also inefficient or he/she wants to give support. In the following we intro-
even wrong. All these plans are put in the same structure, duce each category and describe for each category one
so the task-plan is able to integrate different solutions example of a tool developed.
for the task together with known user mistakes. Plans
can consist of sets of sub-tasks and have a control struc- 3.1. Controlling the course and information space
ture. The decomposition of a plan halts at the time
"posed" primitive tasks are encountered. These primitive An important aspect of the learning environment is
tasks are forming end nodes (leafs) in the tree structure how well it supports the individual learner to access the
and correspond to learner actions or simulation state information he/she is interested in. Two main aspects can
changes. be distinguished. First, how well is a course adapted to
the personal background of a learner. Second, how is the
2.2.5. Creating multimedia and simulation material learner supported in his/her control over the course space.
The actual multimedia and simulation material is cre- In SAM for each learner the author can create an individ-
ated in external packages. The author can choose what- ual course structure. The course structure is presented to
ever package he/she wants to use as long as it meets the the learner with the Instructional Plan browser (IP-
minimal requirements for integration in the SAM envi- browser). The IP-browser (Fig. 3) behaves like an active
ronment and as long as the actual integration is finished. table of contents. It enables the learner to control his/her
The requirements concern acceptance of external control individual course space. It will contain functions such as
and support of dynamic data communication. At the same zoom, run, open, close and search. The IP-browser dis-
time already implemented multimedia material must be plays the learner's progress with the status symbols; seen,
approachable for ISOs. At execution time this multimedia not seen and finished.
material must return the control to SAM whenever it is
finished and it must allow the learner to jump to the 3.2. Understanding the subject and retaining that
browser providing an overview and a free walk through knowledge
the course space. The linking of course items and multi-
media learning material is done with the use of the ISO- The amount of information that the human mind can
editor. process at any given moment is limited. As a consequence
 P. van Rosmalen, J. Hensgens/ Materials Science and EngineeringA199 (1995) 109-120 115

vestigate the properties of the real world by means of

!l-J Instructional Plan Browser experimenting with a model. To do this is not a simple
[] Apprentice boiler mechanic task. Therefore it is useful to support the learner with
~]~[] Understand working of boiler additional tools that support exploratory learning. The
[] Teach working of parts
[] Heater tools should enable the learner to build his/her personal
[] Thermostat knowledge base. The approach follows the classical sci-
t [] Install a boiler entific cycle of exploration, hypothesis, experiment, pre-
diction and conclusion. One example of a tool is the Pre-
Search for: dictions pad. It gives the learner the opportunity to spec-
i[Teach working of parts ify in advance the expected outcome of an experiment
and to compare it at a later stage with the real outcome.

3.5. Communication
Fig. 3. An IP-browserfor a course in boiler mechanics.
An important aspect of the learning process is the exis-
help is required to ensure a successful process of infor- tence of means for testing the level of the learner's un-
mation gathering and information processing. The tools in derstanding of the subject. The learner must be able to
this category are to be used as (1) an aid to both short and show the author that he has understood the subject by
long-term memory, by copying information and creating successfully completing the tasks required from him and
personal notes for later use; (2) an aid to understanding: also providing the rationale of the process he followed to
by restructuring the learning material in an easier-to- reach that conclusion. A tool for this purpose is the Re-
understand format such as diagrams, tables of data. One port Generator. It provides a structured environment in
example of a tool is the Multimedia Notepad. The Multi- which the learner can write his/her answers and use ele-
media Notepad enables the learner to store parts of infor- ments like numerical results, charts, pictures, sounds, or
mation, write comments, and draft his/her ideas. even movies to better explain and support his/her argu-
ments. The report, once finalised and saved, is accessible
3.3. Analysis and display of simulation data to the author and other users, as a reference or for ex-
change of ideas.
An important aspect for the learner is the possibility to
analyse and display the results of series of simulation 4. Specifying a SAM configuration
runs. The tools in this category are to be used for data
analysis and graphical display of the data. An example is The SAM platform is a multi-application framework,
the Analysis tool (Fig. 4). A basic as well as an advanced which will be used to develop simulation-based courses.
level of spreadsheet functionality is provided through this A SAM course can be developed to teach a specific
tool. Various number formats can be stored and manipu- simulation and its domain or it can use one or more
lated, and a wide range of numerical analysis functions, simulations to highlight important concepts. The frame-
performing simple or complicated mathematical calcula- work consists of several packages from the shelf, which
tions, statistical analysis and basic databases are at the cover important parts of the requested functionality (see
learner's disposal. Fig. 5 for an example of a SAM environment). These
packages are dedicated simulation environments, multi-
3.4. Exploratory learning media editors, hypertext and computer based training
tools. The philosophy behind the SAM project is to use
A simulation enables the learner to, for example, in- this provided functionality as an existing base for con-

: ~ Analysis Tools

Histogram
Moving Average
Random Number Generation
I E oK II
[ Cancel)
Rank and Percentile
Regression
Sampling I Xe,p )
t-Test: Paired Two-Sample for Means
t-Test: Two-Sample Assuming Equal Variances

Fig. 4. In AnalysisTools the learner can select the appropriate numerical analysis.
116 P. van Rosmalen, J. Hensgens I Materials Science and Engineering A199 (1995) 109-120

COURSE- SAM SIMU-

WARE LATION
. . ~ . . ..... : . ~ . •

•:.:.: .r~.~. : : : ..'7. :-'7 : . : 1 , 1 ~ .

,1'.'.'-'-" '.'.'.'.'.'.'.'.'. ~-'.'.'.'.'.'.'.'1
t -~imula~,-m:
LabV IEW
HypexCaM |
........... f. . . . . . . 11-..

- i •i Ileir-ii)leili - . . . . . . . i.l,l~. "

Learner Tools

Fig. 5. An example of an existing SAM environment. A combinationof of-the-shelfapplications, HyperCardand Labview, and SAM developed mod-
ules.

structing the simulation-based learning environment. At tween the different tools is minimised by using the Con-
the same time we want to enable a user-friendly adapta- troller as a switchboard. This covers the second "few in-
tion of this environment to meet additional requirements terface" principle which must be followed to ensure
like selection of the best suited simulation package for a proper modularity [22]. Other principles, like "Small In-
specific domain or to meet personal preferences like an terfaces", "Explicit Interfaces" and "Information Hiding"
expertise on a specific authoring tool. The IT-market are also followed in the SAM-architecture. This means
evolves very fast. Adaptation also makes it possible to that the tools can be considered as black boxes. Each tool
exchange the tools (authoring, simulation, learner tools can be exchanged by another one, as long as the new tool
etc.) for competing products and new updates. meets the strict interface requirements of the exchanged
Extendibility and interchangeability are important ob- tool.
jectives in SAM. The SAM platform has an open struc- The Controller forms the central module of the whole
ture. Not only external packages but also the SAM devel- platform. The SAM controller takes care of exchanging
oped author and learner tools like the ISO-editor, Multi- data and control flow between the different parts of SAM
media Notepad, etc. are potential candidates for replace- as a running machine, e.g. course runner, ISO runner and
ment. Distinctions between SAM developed and external monitor and its integrated external tools: e.g. courseware,
packages are not absolutely defined. During the life cycle simulation and learner tools. As such it plays a pivotal
of the current SAM project it appeared that some needed role in the SAM environment both at design and execu-
functionality is not (yet) covered by existing software. tion time. However its own functionality is limited, it acts
The chosen option for programming it ourselves (in a as a telephone switchboard handling requests from one
distinct module) was then inevitable. Borders drawn be- tool (client) by means of services from another tool
tween the SAM developed tools and existing software (server). Authors are supposed to be free to define their
packages are based on pragmatic reasons. We have been personal SAM environment before starting to use SAM.
keen on an explicit definition of the inter-application in- They make the selection, that is most appropriate to their
terface for each module to keep the possibility open for
future exchange.

4.1. The S A M A r c h i t e c t u r e

The SAM architecture is based on the assumption that

it is possible with the current state of the major operating
systems to communicate between different applications
running simultaneously (see Ref. [21] for a discussion of
the Federated Approach, an approach very similar to the
SAM one). It requires inter-application communication
between all modules. The different tools are seen as
(high-level) objects communicating with each other by
using sets of exported services. All tools are autonomous, Fig. 6. All inter-application interfaces are directed via the Controller.
coherent and organised in a robust architecture embodied SAM can be extended with any tool-x, if it provides the required ge-
in the Controller (Fig. 6). The number of interfaces be- neric interface.
 P. van Rosmalen, J. Hensgens I Materials Science and Engineering A199 (1995) 109-120 117
.....................................................,...

i. :. ;- ~iiiiiiiiiiiiiiiiii~iiiiiiiiiiiiiiiiiiii!iiii -i

:J
Fig. 7. Two examples of existing SAM environments. At the left the author uses different tools, i.e. IP-editor, ISO-editorand HyperCard, to support
distinct authoring steps and at the right the author uses only one authoring tool, Authorware, for the same authoring steps.

wishes and experiences, out of a list of SAM-compatible fied location extends the SAM-environment automati-
applications. cally. Whenever a SAM configuration is started, the
Controller finds all external tools and integrates them.
4.2. Creating a personal instance o f S A M This means that SAM makes use of dynamic integration.
The extension with a new external tool has no effects on
In Fig. 7 an example is given of two different SAM other tools. Of course new tools will offer extended func-
environments. Creating a personal instance of the SAM tionality for authors and learners, which can be scripted in
environment is a complex task in the case where an the ISO-editor.
author wants to use a new, not previously integrated tool.
However, the set of additional requirements for a package 5. An example course in SAM
that allows technical integration is kept minimal, to make
an exchange and/or extension practically possible. The SAM has been evaluated by constructing and using
open architecture of SAM is based on the assumption that simulations in two different domains: socio-economic and
it is possible with the current state of the major operating engineering. In the socio-economic domain, we use a
systems to communicate between different applications model of an island. The model [25] was developed for
that run simultaneously. Protocols exist for generalised UNEP (United Nations Environment Programme) to
inter-application messaging. We use Dynamic Data Ex- study the impact of climatic change on the socio-
change (DDE) of MS-Windows [23] and AppleEvents of economic and socio-ecological systems of small island
the Macintosh [24] under System 7. In this way informa- states in the Wider Caribbean Region.
tion can be passed directly from one application to an- The model consists of two levels: at one level the so-
other. The mechanism is appropriate for sending mes- cio-economic growth of the island is modelled (macro-
sages and data and to handle data arriving. scale dynamics), at the other level this growth is distrib-
To integrate a tool in SAM the author has to go uted over the small cells the island consists of (micro-
through two steps. (Note: in the current prototype these scale dynamics). The two levels are clearly recognisable
steps have to be performed by a member of the SAM de- in the interface of the island application (Fig. 8). All ele-
velopment team. This, to assure that the process is done ments of both levels can be made accessible. However, in
correctly and because not all necessary details have been the course being made we will concentrate on population
accurately documented.) First, he/she has to check that dynamics, i.e. the social sub-system of the macro-scale
the selected application sufficiently supports DDE (for a dynamics.
80386 + PC) or Apple Events (for a MAC). Second, he has The SAM set-up in which the course is built, consists
to supply SAM with the necessary information to allow at the moment of the following modules. Authorware
the use of the communication protocols. (Authorware® Professional TM 2.0) as an authoring pack-
To hide from the author the complexity of these proto- age, the island model as the simulation, Excel (Microsoft
cols and the package dependent technical details, the fol- Excel 4.0) as a learner tool (a prediction analyser), the
lowing two lists must be defined for each application, ISO-module to create and run ISOs, and in the back-
namely: ground the SAM controller to take care of all communi-
- a list of commands providing the external control cation and control between the modules. The set-up runs
from within SAM; on a 80486 PC under MS-Windows 3.1.
- a list of all accessible variables of the package with So far the course contains a theoretical part and an ex-
their internal names. ercise. The theoretical part consists of a sequence of
The resulting information has to be saved in a Control screens explaining the principles of population dynamics
& Communication (C&C) table for each integrated tool. applied in the simulation. It was made with Authorware
The positioning of a newly created C&C table on a speci- (Fig. 9). The exercise aims to test the learner's u n d e r -
 118 P. van Rosmalen, J. Hensgens I Materials Science and Engineering A199 (1995) 109-120

Jill) "Fools Oplions Itelp

1~ N e U r a l !:!:i~::..:.:.~!~!:!.~.::!.~.:::.~:!:~i:i:.:~i:.::i:~:i:.::.::i:i~:[:i~:!~!:!:i~:i:!:~::
i:~!i:.":i ~~~ii:-:~:i!i!iiii~!i:i!i!i!i!iii!iii!iii!i!i~!~!~!:i:i:!~iiii!i!i!iii..F~.c~i:~.~:~i!~
::- 1,;1,1 ,;,1,,,:,;.;.:.:.:.:.:.:.:.:.:.;.;.;.:.;.;.;.:.:.;.;.:."
:::: i:i¢~i:i:i:i:i:i:i:i:i:i:~:i:i:i:i:i:i:i:i:i:i:!:i
::1: .~. :::::::::::::::::::::::::::::::::::::::::::::::::::::::
,o,.,. ,,.,.,.,.,.,.,.,::, ..............r.....-.......... . . . . . . . .
[] Lot., A,, iiii!iii~i!i~i!iiii-!.:.-'..'.:ii:..:!i!!!'!.:.:!!..'.:i!i!i!i!i!i!i~~ ~!~!! ~: ............................~:~
II ~!~i~i
~...:~:~:~.~*~:~....~.::~...~:~.~:~.~.~.:~?~:~?.:..~.~:::::::::::::::::::::::::::::::::::::::::::: ~'r;--',;,-',:':::
! ~ i : ~,,:~-:: ~o,., ~ . - , , - - I-:':':':".'-:':-:'"
~:~:~;~i~
[ ] Mousing
,~,~,,,. . . . i.?i-i.'i.'-:~~ ...................... -;~'~ ~ i ~ U i i i ! i ! i ! - i ! i i i t i i ! ~ . . : ' : . - ~ I
[] ~. :::::-:~::~~ !:;:::~::: .... ~!::~ ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
..........: : ...........................

-" :,:7:: i* "i :iiiiiiiii

~ , * : *iiiiiii
: * : , !~i*i* : , : : *i : : , ' : : , : : : ~ : ', ::-::::::::::~.::-::::~-~:.::-~:~-::~:::.:::::::
....... :::::::::::::::::::: :::: ::::: ::::::::::::.,. ,:: ~ ~:-::: : ~ : : ~:. . ~: " :: : -
::::::::::::::::;::::::::::::;;;;m~;~;~;~;~;~:~r.~- ,
m ~i,poa

Fig. 8. The interface of the island simulation. At the top is the Control Panel and the menu-barwith which you control the simulation. At the left is the
window displaying the micro-scale dynamics, i.e. the land-use on the island and how it evolves in time. At the right is the window depicting a graphi-
cal representationof the macro-scale dynamics. It shows the three sub-systems,their variables (as boxes) and the most important loops (as arrows). By
clicking the boxes you get access to relevant variables and parameters for that part of the model.

standing by asking him/her to make a prediction on the tO an introduction in Authorware. In the introduction
population growth in given circumstances. The exercise the objective of the exercise is explained and the nec-
was defined with the help of the ISO-editor and it makes essary instructions are given.
use of all the modules in this set-up. l n i t i a l i s a t i o n . In this block the author defines the
The author defines a script in the ISO-editor to create initial state of the simulation for this exercise.
the exercise. At runtime the script is executed under the W h i l e ( I t e r a t i o n < 20). In this block the simulation is
control of the learner. The script (Fig. 10) contains the started and for each simulation step (a year) the
following: population is retrieved and displayed in a graph pro-
- I n t r o d u c t i o n . In this item the author defines the link duced by Excel.

Fig. 9. Introductionscreen of a demo course on populationdynamics.

 P. van Rosmalen, J. Hensgens / Materials Science and Engineering A199 (1995) 109-120 119

File [dil _~ew + ~cript VC_indow Help

Selec4ed Obl~ect : IDo one step ml zi,mul,mtiom ~!~t

•~ i Inlr0ductJ0n I
Pfogertiez : I~tm~lio n r -~telltlNumbaf o| SlmlpI
~--~('nitialiselionI N u I ~ I ol Step ~ 1

PfopmwtlPValue :
IOo one -*ep of simulation
1oo °°° o,., o, + o , . + t
itbtl 0°, o, po,o,+o°
~ lWrlte to the Excel Speadshe~t

0,-'+°1

Fig. 10. A script built with the ISO-editor(the MS-Windowsversion). At the left side of the window you see the complete script. At the right you see
the editor of the "Do one step of the simulation".

- G e n e r a t e P r e d i c t i o n . In this block the Prediction of experiments in an onboard rack. In addition, a number

Generator is activated in which the learner can enter of workshops took place with independent, potential
his/her prediction for the next 20 years. The learner SAM users. The objective was to find out the constraints
can do this by direct manipulation in the drawing and the benefits of SAM in general. Moreover particular
area or by typing in the numbers. attention was paid to a number of the tools developed by
- W h i l e ( I t e r a t i o n < 4 0 ) . In this block the simulation is SAM.
resumed and for each simulation step the population The development team found that the operating sys-
is retrieved and displayed in a graph produced by tems used (System 7 and MS-Windows 3) offer the tech-
Excel. The learner can now compare his/her predic- nical capability to run together multiple applications and
tion and the actual simulated values. dynamically exchange information. However, only a few
- E v a l u a t e P r e d i c t i o n . In this item a macro is activated applications fully support all the necessary features. For-
in Excel. The macro compares the prediction with the tunately the number of these applications is growing.
actual simulated values. Based on the outcome of this Moreover applications with source code in a higher pro-
comparison additional feedback can be given. gramming language can be easily adapted.
The users involved in general appreciated the basic
6. Conclusions and future directions concept and functionality provided by SAM. Obviously,
given the prototype status, they raised a lot of suggestions
At the moment four different combinations of existing and questions. Some of them have already been taken into
applications and SAM developed tools exist, two on the account. In particular they stressed the importance of a
Macintosh under System 7 and two on 80386 ÷ PCs under sophisticated error handling system. This to assure that an
MS-Windows 3. Each is based on the SAM communica- error could be traced back to the module that caused it.
tion architecture. By realising the different examples of A number of plans have been developed as a follow
SAM environments, we showed how an author can com- up to the current project. The current SAM environment
pose a SAM environment based on his/her own personal is a prototype, albeit a sophisticated one. This implies
preference or experience. that there is a need for a full scale validation and a
In these SAM examples the development team have number of improvements and/or extensions (e.g. tele-
built examples of simulation-based learning materials. In learning facilities). We intend to validate SAM in a larger
one case the material was built on top of the socio- scale demonstrator project. In addition we recognise the
economic model described in Section 5. In another case importance of the generic part of the SAM architecture
learning material was built on top of a simulation of a for projects outside a learning context, i.e. for multitool
spacecraft equipment rack cooling system with a number systems.
120 P. van Rosmalen, J. Hensgens I Materials Science and Engineering A199 (1995) 109-120

Acknowledgements [9] D. Browne, M. Norman and E. Adhamin, Methods for building

adaptive systems. In D. Browne, P. Totterdell and M. Norman
(eds.), Adaptive User Interfaces, Academic Press, London, 1990.
T h e w o r k d e s c r i b e d in this p a p e r is partly supported by [10] P. Thomas and V. Mital, Hypertext document retrieval and as-
the E C u n d e r the D E L T A p r o g r a m m e in Project D 2 0 1 0 - sembly in legal domains. In A. Monk, D. Diaper and M.D. Harri-
S A M . D E L T A aims to research, d e v e l o p and p r o v i d e a son (eds.), People and Computers VII, Cambridge University
series o f t e c h n o l o g y - b a s e d f l e x i b l e training solutions. Press, Cambridge, UK, 1992.
Ill] R. Wallsgrove (ed.), Instructional planning. In Final Specifica-
S A M is d e v e l o p e d by a c o n s o r t i u m c o m p o s e d o f the fol-
tions for the MATIC System, MATIC-deliverable 3.8.3. DELTA
l o w i n g partners R I K S (NL, m a i n contractor), C R I A / S Project 1009, 1991.
(DK), D o r n i e r G m b H (G), U n i v e r s i t y o f Surrey ( U K ) and [12] A. Koehorst, J. van der Baaren and P. van Rosmalen, Authoring
I N S E A D (F). W e gratefully a c k n o w l e d g e the contribu- planning for educational systems. In Proc. 6th Int. PEG Conf.,
tions o f the f o l l o w i n g colleagues: Kleanthis Papanikas, Instituto per la Matematica Applicata, CNR, Genova, 1991.
[13] A. Koehorst, Gericht leren in een exploratieve omgeving, een case
A l b e r t A n g e h r n , E n v e r Yticesan ( I N S E A D ) , T e r r y Hin-
study voor SOLA. (Directed Learning in an explorative environ-
ton, G e r a l d i n e Voost, R i c h a r d Stables (University o f Sur- ment, a case study for SOLA.) RIKS internal report 92-12-076,
rey), F l e m m i n g R a v n ( C R I A / S ) and J o h n van der 1992 (in Dutch).
Baaren, A n d r e K o e h o r s t , S e r g e W a r g n i e s and S t e v e n [14] P. Goodyear, M. Njoo, H. Hijne and J.J.A. van Berkum, Learning
W e l l i n k (RIKS). T h e c o n t e n t o f this paper is based on the processes, learner attributes and simulations, Educ. Comput., 6
(1991) 263-304.
d e l i v e r a b l e " S A M d e s i g n and specification v e r s i o n 2"
[15] J. Hensgens, P. van Rosmalen and J. van der Baaren, Simulations
[26]. in courseware. In Proc. 7th Int. PEG Conf. 93, Heriot-Watt Uni-
versity, Edinburgh, 1993, pp. 210-223.
References [16] AAT, Final Report on Advanced Authoring Tools, DELTA Proj-
ect 1010, 1991.
[17] P. van Rosmalen, SAM, simulation and multimedia. In T. de Jong
l 1] J.L. Alty, Multimedia - What is It and How do we Exploit It? In and L. Sarti (eds.), Design and Production of Multimedia and
D. Diaper and N. Hammond (eds.), People and Computers VI, Simulation-Based Learning Material, Kluwer, Dordrecht, 1994.
Cambridge University Press, Cambridge, 1991. [18] Wallsgrove, R. (ed.), Interpretation of complex learner action
[2] P.M. Greenfield, Electronic technologies education, and cognitive patterns. In Final Specifications.for the MATIC System, MATIC-
development. In D.E. Berger, P. Pezdek and W.P. Banks (eds.), deliverable 3.8.3. DELTA Project 1009, 1991.
Applications of Cognitive Psychology: Problem Solving, Educa- [19] F. Ravn, ITSIE Monitor Taxonomy, CRI/ITSIE/016. ESPRIT
tion, and Computing. Lawrence Erlbaum, Hillsdale, N J, 1987. Project 2615, October, 1991.
[3] L. Plowman, An ethnographic approach to analysing navigation [20] T.P. Moran, The Command Language Grammar: a representation
and task structure in interactive multimedia: some design issues for the user interface of interactive computer systems, Int. J. Man-
for group use. In A. Monk, D. Diaper and M.D. Harrison (eds.), Machine Stud., 15 (1981) 3-50.
People and Computers VII. Cambridge University Press, Cam- [21] R. Riemann Hershey and D. Craig Whitehead, The Federated GIS
bridge, UK, 1992. - A Macintosh Approach. In Proc. 2nd European Conf. on Geo-
[4] D.M. Laurillard, Computers and the emancipation of students: graphical Information Systems, Brussels, Belgium, 1991.
giving control to the learner, Instructional Sci., 16 (1987) 3-18. [22] B. Meyer, Object-Oriented Software Construction, Prentice Hall,
[5] E. Soloway, Ways of seeing, Commun. ACM, 37 (2) (1994) 15- New York, 1988.
20. [23] Microsoft Software Development Kit: Guide to Programming,
[6] E. Rivlin, R. Botagogo and B. Shneiderman, Navigating in hyper- 1990.
space: designing a structure-based toolbox, Commun. ACM, 37 [24] Apple Computer, Inc., System 7 Reference Manual, Apple Com-
(2) (1994) 87-96. puter, Inc., 1991.
[7] K. Tait et al., S1M/10 First Intermediate Report on authoring tools [25] G. Engelen, R. White and I. Uljee, Exploratory modelling of
for simulations, DELTA project DI014 SAFE, 1989. socio-economic impacts of climatic change. In G.A. Maul (ed.),
[8] J.J.A. van Berkum, M.J. Byard and K. Tait, Formalisation of Climatic Change in the lntra-Americas Sea, Edward Arnold,
instructional support in intelligent simulation learning environ- London, 1993.
ments. In T. de Jong and K. Tait (eds.), Towards Formalising the [26] P. van Rosmalen (ed.), SAM Design and Specification Version 2.
Components of an Intelligent Simulation Learning Environment. Report to the European Commission DELTA Project D2010,
DELTA Project 1014, 1991. 1994.