The document discusses embedding physics simulation models created in SimulationX into virtual reality training simulators. It describes how SimulationX models of electrical switchgear were exported as Functional Mock-up Units (FMUs) and connected to an interactive virtual reality (IVR) environment using the Functional Mock-up Interface (FMI) standard. This allows the detailed engineering simulations in SimulationX to drive the visuals and interactivity in the IVR training simulator. The approach provides an engaging training experience with realistic system behavior and responses.

1. Embedding SimulationX Models into Virtual-Reality
Training Simulators of Power Generation Plants
Steve Pantony and Tareq Fityani
Aggreko International, Dubai, UAE
Francis Marinho
Skills2learn, Milton Keynes, UK
Andreas Abel
ITI, Dresden, Germany
Kurzfassung
Die Verbindung von physikalisch und funktional korrekten Simulationsmodellen mit
einer Visualisierung und Interaktion in virtueller Realität (VR) ist attraktiv für Ausbildungs- und Trainingsanwendungen, insbesondere wenn das Verständnis von Bedienhandlungen und Reaktionen einer Anlage wesentliches Ausbildungsziel sind. Dies
gilt umso mehr, wenn das Training an den realen Anlagen kostenintensiv oder bei
Fehlbedienungen potenziell gefährlich ist. Die Umsetzung einer solchen Lösung ist
u.U. schwierig, da VR-Werkzeuge in der Regel keine Modellierungs- und Simulationswerkzeuge bereitstellen und physikalische Simulationstools nur unzureichende Visualisierungs- und Interaktionsmöglichkeiten bieten. Bei Aggreko wurde eine Lösung
etabliert, die dieses Ziel durch die Verbindung von SimulationX-Modellen mit einer
Interaktiven VR-Umgebung (IVR) über Functional Mockup Interface (FMI) erreicht.
Abstract
The combination of a physically and functionally correct simulation with modern
virtual reality (VR) visualization and interaction has a strong appeal in applications
where the operation and respective response of equipment needs to be thoroughly
understood by the operators. This is especially true when training on real equipment
is costly or potentially hazardous in the case of operating errors. The achievement of
this goal is not necessarily straightforward since established VR tools lack powerful
and flexible simulation engines, whereas typical simulation tools have only limited
visualization and interaction capabilities. At Aggreko this gap has been successfully
closed and a process has been established, where SimulationX simulation models are
embedded into an interactive virtual reality environment (IVR) using the Functional
Mockup Interface (FMI).
267

2. 7 SimulationX in Education and Training
Introduction
Among other disciplines, technical training is in need for simulation based applications. The need arises from the fact that due to safety and cost limitations it is not
always possible to train technical personnel on the actual equipment. The aim of a
simulation application would be to deliver accurate, and often complex, engineering
calculations in a visually engaging manner. Due to logistical and efficiency of code
maintenance reasons, another desirable feature of the application would be to have
instances of the simulation engine running in a centralized location (server) with the
possibility of connecting remotely from another location to run the application. All of
the above served as the driving force behind the search for the optimum solution that
would address all the issues. Finally this was realized with the combination of SimulationX, FMI, and an Interactive Virtual Reality (IVR) engine.
At Aggreko the approach so far has been applied for training simulations on the operating procedures of gensets (units consisting of combustion engines and generators)
and high voltage (HV) electrical switch gear. The implementation of the latter is discussed in this paper.
Modelling of HV Electrical Switch Gear using SimulationX
The ABB ZS1 switchgear is the standard equipment used by Aggreko for HV switching. The equipment’s highly hazardous operation and frequent use, among other
reasons, encouraged its modelling into a simulated training environment.
Figure 1: ZS1 Virtual Model
268

3. Virtual-Reality Training Simulators of Power Generation Plants
In general, the two main functions of switchgear are isolation and protection. Isolation is realized through a circuit breaker, which acts as an On/Off switch between
two parts of a circuit. The protection of the equipment is governed by an electronic
unit which ensures that all operational parameters stay within limit. If any fault occurs
the electronic unit will command the circuit breaker to the open position to stop the
current flow. The control unit also provides some means of controlling the equipment through user selected commands. Finally, an earthing function is provided in the
equipment to allow the discharge of any accumulated charge in order for the technical
personnel to access the equipment safely. A typical ZS1 cabinet is composed mainly
of the three components mentioned above (however there are exceptions). A typical
Aggreko ZS1 assembly contains seven such cabinets. A common bus-bar connects all
cabinets together thus enabling current flow.
Other components (e.g. the compartment doors) and functions (e.g. sliding the circuit
breaker out) complement the function of the main components but are also governed
by conditions that have been included into the model.
The first stage of modelling the equipment would be to determine all the desired functionality and behaviours the model should contain. Due to the considerable amount
of functionality and behaviour, it seemed advantageous to adopt the incremental
development model.
Figure 2: Incremental Model
In such a model, the total functionality and behaviour would be incorporated at different intervals. The first version would contain basic functionality and will then go
through the whole cycle of development. For the subsequent version additional functionality will be incorporated on top of previous version.
After formulating the requirements of a version, an exact specification, which will
serve as a reference, of what functionality and behaviour the model should contain
would follow. Also, the inputs and outputs of the FMU would be specified at this
stage. It is highly desirable to specify a list of I/O that takes into consideration all
functionality in both current and future revisions to avoid modifying it at a later stage.
This will allow in certain cases modifying the FMU or even replacing it with another
version without having to modify the frontend.
269

4. 7 SimulationX in Education and Training
Once the specifications are formulated, a plan on how to achieve the intended model
will be established. At this stage a rough algorithm describing the logic of the model
will be written to help guide the development phase. Also, the selection of the appropriate modelling tools would take place during this stage. In the case of the ZS1 the
State Chart Designer in Simulation X was selected. This type of design was found to
be suitable for modelling the logical behaviour of the equipment as it was intuitive to
map the various states the actual equipment can take to their equivalent states in the
state machine.
Figure 3: Behavioural State Chart of ZS1 in SimulationX
Testing is performed in two stages. In the first stage, the model is tested in the Simulation X environment. Generally, this would reveal explicit errors in the design (e.g.
Syntax errors, logical errors, etc.).Once the first test is passed the model would then
be exported into an FMU (Functional Mock-up Unit). The FMU will then be tested to
reveal any potential interfacing issues that might have not been detected in the previous test. In case any faults are detected (in either stage) the process will go back to
the development stage and faults will be rectified and testing will be performed again.
FMU and Code Export
The Functional Mockup Interface (FMI) [1] is a joint industrial effort for exchanging
dynamic simulation models in a unified and standardized manner. It is independent of
tool exporting a model and does not require tailoring models to a target environment.
The exchanged objects – Functional Mockup Units (FMU) - are containers, which carry all data necessary for embedding and running the model. These are in particular the
source code in C, binaries for individual platforms and an XML description of inputs,
outputs and parameters of the model. The calling conventions for all functions of an
FMU are defined and published in the FMI standard. This permits a high degree of
automation when embedding such a unit in the target environment without a need
to adapt the model code during the process. Consequently, all involved parties can
proceed in their developments independently and collaborative work can focus on
270

5. Virtual-Reality Training Simulators of Power Generation Plants
just the definition of interfaces (input and output signals, parameters of the FMU).
Such definition has been the starting point of the collaboration between Aggreko,
Skills2learn and ITI and meanwhile has been standardized through the Simulation
Design Document discussed in the previous section.
FMU exist in two variants – with and without a differential equation solver. Here FMU
with solver have been selected in order to transfer the SimulationX solver capabilities
together with the model to the IVR.
The export of simulation models as FMU is a standard SimulationX feature. The process is assisted through a graphical frontend, where I/O and parameters are defined
per drag and drop from the model hierarchy and the compilation and packaging of
the FMU container is executed automatically.
IVR Training Environment
The interactive virtual reality (IVR) training environment is a highly engaging and
immersive setting that allows the learner to experience real life situations in a controlled and safe manner, creating conducive conditions for information absorption
and knowledge transfer. Experience shows that learning quality and speed improves
significantly when the learner is suitably engaged, through highly visual, audible and
interactive methods. These features are exactly what the IVR environment relies on,
and what makes it so successful as a training vehicle.
The main issue with the detailed engineering simulation models afforded by the likes
of SimulationX for learning is that they target a very low level due to the understandable preferential focus on the complexity of engineering principles over limited visualization. As a result, they cannot be easily explained or understood in non-technical
environments without a detailed knowledge of the underlying functionality.
The use of modern modelling tools and game engines allows the creation of highly
visual environments representative of the real equivalents which gives the learner a similar experience in a controllable / configurable environment. While these tools focus
on the visualization, and have elements of physics and engineering systems in place,
they cannot easily be used to model the same level of detail or complexity of physics
and engineering system simulations as tools such as SimulationX.
Merging the two technologies is therefore of huge benefit as the resulting solution
will exhibit the best of both worlds i.e. a highly visual and interactive environment
with realistic engineering system calculations driving the physics, logic and information. This approach is made possible by the creation of an FMU/FMI, an interface
which contains the functionality, inputs and outputs of a given engineering simulation. This interface allows the IVR to connect to it and control the simulation via input
updates (from user interactions for example) which then get processed and generate
corresponding outputs. These outputs can then be used to update information displays and states within the IVR.
This approach utilizes the strengths of each technology to create a more effective learning tool. It also allows development strengths to remain focussed on their areas i.e.
271

6. 7 SimulationX in Education and Training
the IVR developers can focus on getting the environment as visually rich as possible
without worrying too much about what goes on behind the scenes, and the simulation developers can focus on getting the engineering / physics functionally accurate
without worrying about how it’s presented.
Figure 4: IVR/FMU workflow
The most effective and scalable delivery platform for anything these days is the Internet, and as such, the ability to have a learning tool which is highly visual, contains
realistic engineering / physics processing, is accessible from any capable machine that
has access to the internet, and can be integrated into learner management systems
or virtual learning environments is of immense value to any training provider. The
additional advantage of web based learning tools is that the content can be updated
in one place and all learners who have access to the content can instantly be using
the latest version.
Figure 5: IVR/FMU Web design workflow
Using the FMU/FMI over the internet has the added benefit of security of the content
and means that it doesn’t need to be stored directly with the learning tool, which can
make maintenance and updates easier. It also improves performance to a degree, as
the overhead of running the FMU is offloaded to a separate machine from the one
running the training material. The web technologies used include:
•• Web sockets for passing messages between the client and the FMU server, and
between the FMU server and the FMU
•• A windows based server for handling client requests
The FMU server currently exists as a .NET application on a Windows server with communication being handled by TCP sockets. The server can handle multiple different
272

7. Virtual-Reality Training Simulators of Power Generation Plants
FMU models simultaneously, with multiple clients connecting to them. The future
development plans include:
•• Hosting the FMU/FMI and FMU Server on the cloud. By utilising a cloud infrastructure, the load can be distributed on a scalable platform and be more
accessible from anywhere in the world.
•• Enhancing the FMU server for increased compatibility with a wider variety of
platforms
•• Bringing the server in line with the next incarnation of FMI standard (FMI 2.0)
•• Implementation of a web based control panel to allow easier management of
FMU’s, connected clients, restarting of the server etc.
•• Implementation within an Mbook allowing the content to be accessible via
mobile devices
Model Embedding
The basic process of embedding the FMU model consists of establishing a connection
to the FMU Server, which then initialises and loads the relevant FMU using the functionality dictated by the FMI.
The FMU contains predefined inputs and outputs which the IVR is aware of and it
initialises the environment by detecting any changed outputs. Once the environment
is adjusted to reflect the start state, via the FMU, it can then be used as normal, with
the user interacting with the IVR elements and receiving visual feedback representations of the FMU outputs.
As illustrated in the earlier diagrams, the client IVR programme sits within a browser
window and knows of the server address that hosts the FMU and FMU server. Once
the FMU server is started on the host server, it waits for a connection from an incoming client.
Armed with the server address and the FMU name that is required, the client establishes a connection to the FMU server using the TCP sockets informing it of the FMU it
needs to use. The FMU server creates an instance of the desired FMU and the initialisation process is complete.
With the connection established, and the FMU initialised, the IVR can then be used
to navigate and interact as it is designed to. With every interaction such as a click on
a door, the action can be sent via the FMU server as an input to the FMU which does
the processing to establish whether the door can be opened in its current state (given
interlocks, circuit breakers racked states etc.). The FMU then updates the outputs
which are sent back to the client via the FMU server. The client receives notification of
the updated outputs and adjusts the visual environment accordingly.
All of the above has been achieved via a hybrid of technologies, frameworks and
programming languages including C# (Mono and .NET), C++, Javascript, HTML, Json
to name a few.
The IVR delivery can be standalone (i.e. run as a local programme via an executable)
or as a web based package (i.e. run by accessing a URL in a browser). Standalone
273

8. 7 SimulationX in Education and Training
programmes can be much larger and computer intensive as they have the full power
of the PC at their disposal, however web based programmes are restricted by the
browser within which they run. Irrespective of the delivery type, the approach used
for embedding the model is identical.
There have been several challenges encountered during our development to date,
which we have eventually been able to overcome. Some of these are identified below.
One of the big challenges was our IVR environment being incompatible with DLL’s
when run in web player mode. This has since been overcome by moving the FMU to
a remote location so that it is not actually embedded directly into the IVR.
Another challenge from the server side was a way to handle premature disconnections of clients. By default, sockets have no real sense of whether or not they are still
connected; making it difficult to work out if a client has disappeared. This was worked
around by having the client and server “pinging” each other. If either side goes more
than a few seconds without a ping from the other, it is assumed that the connection
has been lost.
Conclusions
In this paper we have shown, that the embedding of physical and functional simulation models into IVR training simulators is feasible. Despite the yet small amount of
applications that have been developed using this setup, the potential seems very promising and the process is starting to take form. Standards are being set to facilitate the
growth of the new methodology. This includes, e.g., a design document which acts as
a common reference between involved parties and allows an effective collaboration
between IVR developers, modellers and the end users of the training application.
Hosting the IVR environment on a server and deploying the training contents through
the web allows a unified training approach applied in a decentralized manner even in
large companies and organisations.
The authors are confident that this will only be the start of a new breed of virtual
environments that can be found useful in many disciplines.
References
[1]
274
The Functional Mockup Interface – www.fmi-standard.org