Digital Twin Development
Digital Twin Development
Michael Sauer
Fluid Machinery and Measurement, University of Applied Sciences
Saarbrücken, Saarbrücken, Saarland, Germany
Tuomo Liimatainen
Mechanical Engineering, LAB University of Applied Sciences,
Lappeenranta, Etelä-Suomi, Finland
Dirk Hübner
Lightweight Construction, University of Applied Sciences Saarbrücken,
Saarbrücken, Saarland, Germany
© The Editor(s) (if applicable) and The Author(s), under exclusive license
to Springer Nature Switzerland AG 2023
This work is subject to copyright. All rights are solely and exclusively
licensed by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any
other physical way, and transmission or information storage and retrieval,
electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_1
1. Introduction
Frank U. Rückert1 , Michael Sauer2 , Tuomo Liimatainen3 and
Dirk Hübner4
(1) Fluid Energy Machines, University of Applied Sciences Saarbrücken,
Saarbrücken, Saarland, Germany
(2) Fluid Machinery and Measurement, University of Applied Sciences
Saarbrücken, Saarbrücken, Saarland, Germany
(3) Mechanical Engineering, LAB University of Applied Sciences,
Lappeenranta, Etelä-Suomi, Finland
(4) Lightweight Construction, University of Applied Sciences
Saarbrücken, Saarbrücken, Saarland, Germany
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Fig. 1.1 Sketch after a scene from petroglyphs in Bryce Canyon National Park, USA [1]
What we don’t see in Fig. 1.1 is that there are connections between the
different objects. They are not painted, but in fact, there is a story and a
connection between the different symbols and people. Two larger persons
with antlers on their heads could also be identified as supervisors evaluating
the situation and the quality of the hunting scene. We want to demonstrate
the idea behind simulation models for problem-solving processes and how
we can create artificial twins for any kind of problem. The core problem is
that today’s technical solutions are becoming more and more complicated.
More details are included, leading to unexpected behavior. Simulation
models can help understand your work, but—beware—they can also make
it worse because they make the system more complex. Too much ingenuity
can very quickly lead to failure. An engineer has to understand the aim and
purpose of his simulation to answer the question why. After that, answer to
question how is often more trivial. At the end of this book, in the
Conclusions chapter, we will discuss further methods for teaching and
introduce an additional new idea on how to teach complex technical issues
very quickly by introducing a role play method.
During creating the so-called digital twins, we wanted to introduce the
simulation tool Simcenter Amesim, which can be used to simulate the
behavior of systems. The software we use is from Siemens PLM Software
[2]. A free-of-charge demo version can be installed without much effort on
Microsoft Windows or Linux systems. Working on Apple operation systems,
one can use the program via a virtual private network (VPN) connection. Of
course, this is less convenient. The software tool contains a ready-made
library, including smaller models to simulate pneumatic, hydraulic, thermal,
electrical, and mechanical behavior. It additionally includes a theoretical
library for signals, mathematics, and artificial intelligence, e.g. artificial
neural networks (ANN) or reduced order modeling (ROM).
We will explain the basic functions and how to start modeling with this
software tool. It is considered as a multi-domain software, meaning that all
the different physical domains can be linked with each other or with the
signal library. The software offers the possibility of simulating systems
without the need for computer-aided design (CAD) geometry, but 3D
geometries can be added and controlled by the models if this is needed.
These models can also be printed with a 3D printer. In former times, small
models made of wood or clay were built to study the behavior of systems.
Today, our students are fascinated by 3D printing, which plays a similar
role. The driver is similar to that of the petroglyphs, and is why we use it for
a clearer understanding.
The usable symbols for the different libraries in Simcenter Amesim are
partly standard symbols defined and therefore easy to recognize. The
software allows it to interact with other programs such as Excel or
MATLAB/Simulink. One big advantage is, if compared, for example, with
Simulink. Many industrial models are already finished and can be used as
reference cases from the onset.
In order to bring the basics closer in the study, it was necessary to plan a
lecture with the contents, e.g. the artificial model construction and the
practical application of the used software. As previously described, the
fascination lies not only in the manual creation of the simulation model but
also in the possibilities that arise when one prototype has to be constructed.
With a digital twin, technical, physical, or biological components can be
tested long before they are used. The digital twin must reproduce certain
properties of its realistic model to allow its functions and behavior to be
scrutinized.
The earlier in the development process the engineer has the opportunity
to make quantitative statements about the function of a component in the
system, the more effective a later design or development process can be.
The digital twin is the image of its real counterpart and represents the first
step in the development process. In our daily work experience, there is a
difference between creating and using a numerical model. Even though the
creation of the twin requires an expert who must know the process to be
mapped down to the smallest detail, the user of the digital twin does not
need to understand everything in detail in order to work with the twin. We
use the example of the light switch. You do not have to understand how it
works to use it. The gap between creation and use should be bridged by this
book. Possibly this was also one idea behind the petroglyph scene in
Fig. 1.1.
The relevance of this topic in today’s world is obvious. Also in the most
diverse areas of product development. Today, the topic of sustainability also
encompasses the behavior of a product over its lifetime cycle. This means
that a simulation model should not only make a statement about how a
technical product works and which functionality in the various phases of its
life but also how large an effect really is and what its consequences are is
only known to a very limited extent. Today, for example, a very small
amount of a powder containing anthrax virus can start a war. Quantities of
harmless gas that are only in the ppm range (ppm = parts per million) can
bring an entire industry to a standstill. On the other hand, sometimes small
amounts of a virus are enough to have a very large effect. In such a case, the
complexity of the overall system is crucial. When we build systems more
complex, they get more susceptible to small quantities.
1.2 What Is a Digital Twin?
When creating the model or digital twin from reality, it should be noted that
the original cannot be reproduced in its entirety. Rather, certain properties
are to be reproduced in isolation, i.e. in digital form as in Fig. 1.2. What is
meant by this? We can explain it by the example of a human body or better
with a prometheus, as described in [3].
In the case of this human being, like in Fig. 1.2, it can be possible to
digitally reproduce only the blood vessels and thus create a kind of
hydraulic twin. Such a twin would not provide any information about the
bone structure. To represent the statics of the bones of the human being, one
would have to create a mechanical twin. The original can therefore have
different digital twin brothers, which only partially reflect the physical or
geometric properties of their real counterpart.
Fig. 1.2 Digital twin is the counterpart to the real twin, but with limited capabilities
Archetype Description
Basic Digital This digital twin provides an human-machine interface (HMI). It therefore
Twin extends the digital twin by a single feature and is therefore considered
fundamental
Enriched Based on basic archetype, this digital twin enriches its database by preprocessed
Digital Twin data from supplementary systems. Many objects in this cluster describe the
possibility of semi-manual data collection
Autonomous This twin is an advancement from enriched digital twin. It offers autonomous
Control Twin control, but at the same time, it contains an HMI for the option to intervene. As
direct communication with another virtual or physical machine, this archetype
needs at least interoperability via a translator interface
Enhanced The digital twin offers autonomous control over a physical asset while integrating
Autonomous external, downstream data processing systems. The interoperability needs to be
Control Twin secured at least over a translator as well. On the downside, this archetype does not
offer an HMI
Exhaustive A digital twin with exhaustive data acquisition options, data processing, and
Twin control over a physical asset. This archetype provides the user with all options.
The twin is able to work and control autonomously. At each point, humans have
the ability to intervene or to enrich the database and, hence, provide a semi-
manual data linkage. This archetype demands a fully interoperable data linkage to
downstream systems as well as to the physical asset itself
Important: You can test the program before you buy it, find a free
student version of Simcenter Amesim at the following link:
https://www.plm.automation.siemens.com/plmapp/education/simcenter/
en_us/free-software/student
Paths for software can change, and books are usually more persistent
than websites on the Internet. Therefore, the specified path may no longer
be correct. In this case, try to find the described software using a common
search engine. The free-of-charge version from Siemens has some
restrictions such as limitation of the library. With this student version, the
tasks were designed for a lecture plan of this textbook. Only the available
libraries are explained in more detail below and all examples could be done
with the free version.
Fig. 1.5 Graphical user interface of Simcenter Amesim offers a clear overview
Mode Description
Sketch mode Drawing the model on the worksheet
Submodel mode Selection of the different submodels from the libraries
Parameter mode Detailed definition of the physical parameters
Simulation mode Run and review the simulation and graph the results
To add physics for the technical problem we have to choose the right
physical library. Each library has its own color. The signal library is usually
illustrated in red color. We want to take the chance to introduce the
functions of Amesim in detail.
Signals and Control library The signal library contains different
mathematical functions and is used to control the other libraries. Basic
mathematical functions are available. We will explain in the next chapter
how to use them. This library contains artificial intelligence models.
Mechanical library This category contains linear and rotary motion
elements. It is often used to complement other library categories to
simulate mechanical ratios. Components such as masses, springs,
dampers, friction elements, and also transmission ratios can be found in
this section.
Hydraulic library This section contains general hydraulic components
such as pumps, pipes, and valves for the simulation of the dynamic
behavior of fluids.
Pneumatic library It contains components that are used to model
compressed air-based systems. Typical components include pressure and
flow sources, valves, or gas properties.
Thermal library With the thermal library, heat states and interactions
can be simulated. The thermal components such as convection,
conduction, or even radiation elements enable calculations of heat flow
and losses.
Electric library In this library are electrical components with which
systems can be reproduced, especially for automotive electronics.
Components are resistors, inductors, capacitors, batteries, and also
already prefabricated motors.
The applications of the presented program are described in more detail
in further sections. The color of the items can also be changed by right-
clicking it with the mouse. But at the beginning, we should leave it in the
default color for better understanding.
References
1. Hiser, D.: Records of the Environmental Protection Agency, The Environmental Protection
Agency’s Program to Photographically Document Subjects of Environmental Concern, 1972–
1977. https://catalog.archives.gov/id/545671
2.
Siemens AG: Simcenter Amesim. https://www.plm.automation.siemens.com/
3.
Shelley Wollstonecraft, M.: Frankenstein, or, The Modern Prometheus. Oxford University Press,
New York (1818)
4.
Hendrik van der Valk, H., Hasse, H., Möller, F., Otto, B.: Archetypes of digital twins. Bus. Inf.
Syst. Eng. 64(3), 375–391 (2022)
5.
Kim, Y.-W.: Digital Twin maturity model, WEB 3D 2020 industrial use cases workshop on digital
twin visualization. https://doi.org/10.13140/RG.2.2.28750.48967 (2020)
6.
Weisberg Business Consulting GmbH: Einfach erklärt: Was ist Big Data? https://weissenberg-
solutions.de/einfach-erklaert-was-ist-big-data/
7.
Smith, S.E.: Neuromuscular junction. In: Zaimis, E. (ed.) Handbook of Experimental
Pharmacology, vol. 42, p. 593. Springer, Heidelberg (1976)
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_2
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Fig. 2.3 Example: Creating a simple calculator; Step 2: Copying constants and adding a sum
function
Fig. 2.4 Example: Creating a simple calculator; Step 3: Connection of the two constants and the
sum function
Fig. 2.5 Connecting an item for constant value with the item for summation
When all ports of a model are properly connected, the color of the items
changes. The model is then no longer highlighted in dark color and can be
used. If the connecting does not work, there is usually a thinking error in
the design. This can be checked if you compare the units of the two sides of
the connections.
Important: Always make sure that the elements are connected correctly
and highlighting has been changed. It is also possible that the elements
cannot be connected if a physical model is illogical, e.g. if the units do
not fit. In this case, you should question if your model is logical.
In general, you can connect the elements from the signal library with all
other libraries. A sink from the signal library, as shown in Fig. 2.6, must
therefore always be compatible with the other libraries and can then be used
accordingly if you cannot find a suitable connection or continuation of the
model for a particular simulation model in a physical library.
Fig. 2.6 Example: Creating a simple calculator; Step 4: Closing the model with a sink item
Once the model is complete for addition, no item will be darkened. All
items should then be displayed in normal color. Now, you can switch from
Sketch mode to Submodel mode as shown in Fig. 2.7.
Switching to the Submodel mode is possible only if the model has been
drawn correctly before. In order to enter individual values in the graphical
model, it is then necessary to switch further to the parameter mode as
shown in Fig. 2.8.
In the mode parameter it is now possible to enter the values to solve our
equation . You can see exactly where to enter the values in
Figs. 2.9 and 2.10. Additionally, in Figs. 2.11 and 2.12, you can see how to
switch to simulation mode and then read the result of the calculation.
Fig. 2.7 Example: Creating a simple calculator; Step 5: Change from Sketch mode to Submodel
mode
Fig. 2.8 Example: Creating a simple calculator; Step 6: Change from Submodel mode to Parameter
mode
Fig. 2.9 Example: Creating a simple calculator; Step 7: Entering a value for the first constant
Fig. 2.10 Example: Creating a simple calculator; Step 8: Entering a value for the second constant
Fig. 2.11 Example: Creating a simple calculator; Step 9: Switching to Simulation mode and starting
the simulation
Fig. 2.12 Example: Creating a simple calculator; Step 10: The simulation is finished and the result
can be read off
Problems
For better understanding, you can change the signal model now and try out
any other calculation [2].
2.3 Copy your model with Ctrl+C, paste it somewhere on the same
workspace, and use two different models in one workspace.
2.4 Plot a diagram of your results. Try to plot two different results from
different models in one diagram.
References
1. Günther, M., Velten, K.: Mathematische Modellbildung und Simulation, Wiley-VCH (2014)
2.
Bossel, H.: Modellbildung und Simulation. Springer Vieweg, Wiesbaden (1992). https://link.
springer.com/book/10.1007/978-3-322-83658-8
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_3
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Fig. 3.1 Mechanical simulation model for a falling ball that is bouncing on the ground
Important: Focus on a correct sketch of the model in the beginning and
try if it can be compiled and the simulation runs. For this, estimate the
physical parameters and boundary conditions first. After that step you
can refer the correct quantities and add them. This will help you to keep
track.
Simulation settings
Start time = 0 [s]
Final time = 4 [s]
Print interval = 0.01 [s]
When the progress bar has risen to 100% as shown in Fig. 3.3, the
simulation was completed successfully. Now the results can be examined. If
a simulation does not run successfully, or if it takes a very long time to
generate a result you have to change the simulation parameters. There is the
possibility to choose the simulation time (Final time) a bit lower and you
could also sets the Print interval to a smaller time step size.
Fig. 3.3 Successfully completed the simulation for the digital twin of a falling ball
This allows you to check which possible error could be present in the
simulation model. Mostly the error is caused by physical values that do not
make sense, e.g. the mass of the ball is much too high or the distance is too
large. If you select the time steps too large, it is possible that the simulation
cannot be carried out successfully. It is then said that the mathematical
model does not converge. But also if you choose the time steps too small it
is possible that the simulation cannot be executed successfully. This is
because the Amesim program writes results to the hard disk for each time
step. If too much data is written, problems with disk space may occur.
Fig. 3.4 Generating diagrams: visualisation of the result by dragging the variable on the work space
Next, let us take a look at the most important simulation results that
result from the simulation for our ball model. We want to analyse them.
Drag the desired result variables over the workspace to create a diagram of
the value as shown in Fig. 3.4. After that, we will give some suggestions for
further developments or further investigations on our digital twin. One thing
should be kept in mind when creating the model. If something would not
work in reality, e.g. if the mass of the ball or the distance to the ground
would be much too high, then this can also lead to an abort and an error
message. That’s why simulations should only be done with halfway
plausible parameters.
Fig. 3.5 Size of gap is a measure of the distance between the ball and the ground
Important: Drag a result variable with the mouse from the menu on the
right side to the workbench to plot a diagram in an x,y-chart.
To get to know the digital twin of your bouncing ball problem a bit
better, let us work on a few more advanced work suggestions. In addition,
let’s take a look at the speed of the bouncing ball. For this we drag the
variable Velocity at port 1 from the variables field onto the workbench to
display this variable in a diagram as well. When the ball is released the
velocity is still small, towards the end it gets bigger and bigger until the ball
hits the ground and is bounced back.
Problems
3.1 Investigate after what time the ball first touches the ground if you set
the distance between the ball and the ground at the beginning from the
height 1.0, 2.0 and 4 [m].
3.2 How does the trajectory of the ball change when you increase the
weight of the ball from 1.0 to 1.5 [kg] and to 2.0 [kg]?
3.3 How does the trajectory of the ball change if you decrease the contact
damping force from 150 to 50 [N/(m/s)]?
3.3 The Mechanical Rocker
Now let’s consider another use case that we can recreate with the
mechanical library. We want to create a twin for a mechanical rocker. On
the right side, we want the bar of the rocker to be pushed down. On the left
side of the seesaw, we want a spring that is attached between the left lever
arm and the bottom. Let’s start right away and build the model of the rocker
as shown in Fig. 3.6.
Item Parameter
RAMP0 Slope = 1 []
XVLC01 Time constant for derivative of displacement = 0.0001 [s]
LML012 Distance port 1 to pivot = 1 [m]
Distance port 2 to pivot = 1 [m]
SPR000A Spring stiffness rate = 1000 [N/m]
Spring force with both displacements zero = 0 [N]
Table 3.4 Simulation time for the simulation model of the mechanical rocker
Simulation settings
Start time = 0 [s]
Final time = 1 [s]
Print interval = 0.01 [s]
Important: Always check your models for plausibility and logic before
you make the models too complicated. Check the dimensions and the
units of your results.
The fact that the right bar is pushed down is shown by the negative
value. If the left bar had a greater length, it would move further up than the
right bar would move down. So we see that both lever arms are moved by 1
[m] in the time 1 [s].
Fig. 3.8 Height by which left lever is lifted up and the right lever arm is pushed down
Fig. 3.9 Angle in degrees by which the beam of the rocker rotates around its axis
We can also look at something else in Fig. 3.9. That is, by what angle
the beam of the rocker is rotated around its axis. This can be evaluated by
the size angular lever position. Here, the seesaw rotates by about 57 . The
example is relatively simple now, but we can do the following suggested
work to make it a little more interesting and explore our mechanical
problem a little further. In the Problems part you find suggestions for
further work.
Problems
Why don’t they do the following investigations with your rocker model?
3.4 How do the results change when we press the seesaw for more than
one second? At what point do the results become nonsensical and why?
3.5 Check how the height to which the weight is pulled changes when the
mass hanging at the bottom of the rope changes. Does the length of the rope
also change during the process?
3.6 What happens if we extend the right arm of the bar, e.g. to 3 [m]?
Interpret your new results.
3.7 What changes if we increase the spring force on the left arm of the
seesaw? Interpret these results as well.
Fig. 3.10 Simulation model for a rope to lift a weight running over a pulley
The symbol with the apple [GROV0] indicates the magnitude of gravity.
On the earth the gravitational acceleration of 9,80665 [m/s ] acts almost
everywhere. If we were on another planet like Mars or Venus, e.g. we
would have to use a different value for gravity. If we were on our
neighboring planet Mars, we would not be able to use this digital twin
because the gravity there is 3.711 [m/s ]. On the moon, the gravity is 1.62
[m/s ].
We can’t change the gravity acceleration arbitrarily in the individual
items, and the apple icon doesn’t affect all models the same. Unfortunately,
however, you can always paste the icon on the desktop, and that’s what we
did, because the apple icon looks very nice. But inside the submodels,
unfortunately, the gravitational constant of the earth is still used in some
places. This should be adapted once in following versions of the program
Amesim. Then digital twins can also be created for other planets.
Here we want to do the calculation for 4.0 [s]. Please make sure that the
simulation time in Table 3.6 is not too long, otherwise there is a risk that
you will pull the whole rope through the pulley. Our model is not designed
for this and no useful results would come out.
Table 3.5 Parameters for the simple rope hoist simulation model
Item Parameter
CONS00 Constant value = −50 [N]
RSHE002A Diameter of the sheave = 500 [mm]
Roping angle = 90 [degree]
MECROPE0 Stiffness of unit length of rope = 1e+06 [N/m]
Viscous friction of unit length of rope = 1000 [N/m/s]
Initial length = 10 [m]
Item Parameter
RL00A Moment of inertia = 1 [kgm**2]
Coefficient of viscous friction = 1 [Nm/(rev/min)]
(All other values are set to 0)
MAS001 Mass = 1 [kg]
GRAV0 Constant gravity value = 9,80665 [m/s/s]
Table 3.6 Simulation time for the simulation model simple wire rope hoist and pulley
Simulation settings
Start time = 0 [s]
Final time = 4 [s]
Print interval = 0.01 [s]
Problems
Now let’s look at what else we can use our digital twin of the rope running
over a pulley for. They can investigate the following work suggestions on
their own using their simulation model:
3.8 What would happen if they double the force with which they pull on
the rope. How fast will the mass be lifted then?
3.9 Check how the height to which the weight is pulled changes when the
mass hanging at the bottom of the rope changes. Does the length of the rope
also change during the process?
3.10 Try to create a simulation model where you run the cable over two
different pulleys. What does this do for you?
3.11 Just remove the apple symbol and see if the model still runs.
Item Parameter
MAS002 (of the car)
Mass = 1500 [kg]
Inclination = 0 [degree]
SPR000A Spring rate = 100000 [N/mm]
Spring force with both displacements zero = 0 [N]
DAM0000 Damping rate = 1000 [N/(m/s)]
MAS002 (of the wheel)
Mass = 35 [kg]
Inclination = 0 [degree]
Item Parameter
SD0000A Spring rate = 100000 [N/m]
Damper rating = 1000 [N/(m/s)]
XVLC01 Time constant for derivative of
Displacement = 0.0001 [s]
UD00 Constant gravity value = 9,80665 [m/s/s]
Number of stages = 2
Cyclic = no
Time at which duty cycle starts = 0 [s]
Output at start of stage 1 = 0 null
Output at end of stage 1 = 0 null
Duration of stage 1 = 2 [s]
Output at start of stage 2 = 0.08 null
Output at end of stage 2 = 0.08 null
Duration of stage 2 = 8 [s]
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.01 [s]
Fig. 3.16 Oscillation of the car cabin in the moving car after step of the sidewalk
At the end of this section you can work on some problems that we have
defined for you. This is a normal and good approach and shows you how
engineers work. First we created a simple model that works and after this
introduction to the problem we will try to improve our car model more and
more until we end up with a good car behavior.
Problems
We have already discussed some problems. Now you can investigate the
following problems for your damper with your simulation model.
3.12 Try to make the damping of the spring system softer to reduce the
oscillation in the cars cabin.
3.13 Change the size of the step and examine, how the damping changes.
3.15 Work through the chapters on hydraulics and pneumatics and also try
to build a hydraulic or pneumatic shock absorber.
3.16 Try to generate a more complicated road profile and chose longer
simulation times.
References
1. Bauer, H. (ed).: Automotive Handbook, 4th edn, Robert Bosch GmbH, ISBN 0-8376-0333-1, p
584 (1996)
2.
https://www.blender.org
3.
https://poly.cam/
4.
Travis, S.: Introduction to Laser Science and Engineering. CRC Press, Taylor (2019)
5.
https://en.wikipedia.org/wiki/STL_(file_format)
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_4
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
In our twin, we have to use a special element [THCD00] for the heat
conduction between the metal bodies. We composed our digital twin from a
total of four different metal bodies [THC00]. To tell the model which
material each item in Fig. 4.1 is made of, we have to use a so-called solid
type index.
Fig. 4.1 Simulation model for heat conduction, the heat is transferred between a body made of iron
and a body made of aluminium
A heat source of 50.0 [W] in [THHS0] is attached to the right side of the
iron block and to the aluminium block, respectively. At the top is the
aluminium block where heat is dissipated by convection and radiation.
Here, too, it is possible to change the air speed in order to influence the heat
transport.
Item Parameter
THHS0 Heat flow rate at port 1 = 50 [W]
Item Parameter
THSD00 Solid type index = 1 [-]
Material definition = pure aluminium (Al) [-]
THSD00 Solid type index = 2 [-]
Material definition = pure iron (Fe) [-]
THC00 Solid type index = 1 [-]
Mass of material = 5 [kg]
THCD00 Contact surface = 1000 [mm**2]
Thermal contact conductance = 1000 [W/mm**2/degC]
THGCV0 Inclination angle = 90 [degree]
Width = 100 [mm]
Length = 150 [mm]
Velocity of the fluid = 4 [m/s]
It has to be noted, that for the estimation of the contact surface for
[THCD00] the volumes of the [THC000] blocks have to be taken into
account. The simulation time in Table 4.2 is chosen to be very long for this
example, because such heat conduction processes can take quite a long time
power source is low and weight is relatively high. Therefore, we adjust the
time step size accordingly to higher time step sizes.
Table 4.2 Simulation time for the model of an aluminium heatsink
Simulation settings
Start time = 0 [s]
Final time = 3600 [s]
Print interval = 0.1 [s]
Problems
Using your digital twin for the heat sink, investigate the following work
suggestions once:
4.1 Change the materials and use copper instead of aluminium, for
example. How does this change the temperatures of the bodies?
4.2 Investigate what happens if you increase the masses of the iron bodies
from 5 to 10 and 15 [kg].
4.3 Set the heat source on the right side from 50 to 100 and 150 [W].
Observe what happens to the temperatures of the metal bodies.
4.4 Try to cool the iron body further by increasing the air velocity on the
right side to 20 [m/s].
4.5 Have a look at the result output for convection. There you will find
different comparison numbers, like the Nusselt number or the Reynolds
number. What do these numbers stand for?
Fig. 4.5 Simulation model for a solar collector and the heat conduction in the adjacent material
4.3.1 Simulation Model
To design the solar collector in Fig. 4.5, it is important to know where it is
located. The location plays a crucial role in its function. A collector located
near Johannesburg in South Africa has much more solar energy available
than a collector located in Paris or Helsinki. Furthermore, the time of day at
which the solar radiation falls on the collector is of decisive importance. In
most places, the radiation is strongest around noon. In Simcenter Amesim, it
is possible to set these environmental conditions using a custom model.
This makes it possible to design the collector exactly for its particular
position. We could also create the collector from single items. But this is not
absolutely necessary. There is already a ready-made model that we can use.
The solar collector should still be connected with elements made of
aluminium, which should symbolize the design and structure of the
collector support and its heat exchangers. Through the contact, it can give
its energy to these elements and thus will heat them up.
Item Parameter
THSD00 Material definition = pure aluminium (Al)
THAMBCOND0 City name = Paris
Year = 2007
Month = December
Day = 16
Hour = 8
Minute = 00
THGCV0 Inclination angle = 90 [degree]
Width = 100 [mm]
Length = 150 [mm]
Velocity of the fluid = 5 [m/s]
Item Parameter
THC000 Solid type index = 1 [-]
Mass of material = 50 [kg] and 20 [kg]
THR03 Equivalent emission factor wall/gas = 1 []
Exchange area = 100 [m**2]
Temperature of the gas = 20 [degC]
THRSOL002 Solar radiation setting mode = using ambient conditions
Exchange area = 1 [m**2]
Adsorption factor = 0.7 []
Equivalent emission factor gas/surface = 0.9 []
Surface inlination = 45 []
Decisive for the absorption of the amount of heat is the area of the
collector and at what angle the collector is inclined to the horizon. We can
specify many different cities all over the world to indicate the position.
However, it is also possible to enter the GPS coordinates of a location. GPS
stands for global positioning system. A system by which signals are sent
from satellites used to show the position on the globe. This makes it
possible to calculate yields that a solar collector field can produce. The
simulation times are given in Table 4.4.
Table 4.4 Simulation time for the simulation model of a solar collector
Simulation settings
Start time = 0 [s]
Final time = 30000 [s]
Print interval = 1 [s]
Fig. 4.8 Radiant power impinging on the surface of the solar collector
Problems
With the solar collector model is completed, you can now perform various
tests and modifications:
4.6 Once, change the weight of the left aluminium bracket to half its
value. What happens? How does the temperature change when they
increase the wind speed from 5.0 to 8.0 [m/s]?
4.7 How does the energy input to the collector change, if we move the
collector to Johannesburg or Helsinki? Select some alternative cities for
simulation.
4.8 Consider a model setup where they can use the collector to heat a
medium such as air or water.
Reference
1. Incropera, F.P., DeWitt, D.P.: Fundamentals of Heat and Mass Transfer, 4th edn. Wiley (1996)
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_5
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.01 [s]
Important: Always pay attention to what liquid you expect, e.g. whether
it is oil, water, or another liquid. Change the liquid properties, density
and viscosity, with the droplet symbol [FP04]. The index for the fluid to
be used must also be set there. This index can be redefined for each
hydraulic component. Several different fluids are possible within one
larger system.
If you look at the fluid properties of the liquid in this experiment, you
will notice that it is not water but an oil, e.g. hydraulic oil, heating oil or
diesel fuel, due to its low density. Here the index of the hydraulic fluid = 0
was selected.
Fig. 5.2 Filling level of the different oil tanks as a result of the analysis
We want to show the filling level from the upper tank and the filling
level from the lower tank in the same diagram and can now compare the
results very well. Since both tanks are exactly the same size, the level in the
upper tank has increased by exactly the amount of oil that is missing in the
lower tank.
Fig. 5.3 Torque of the pump shaft and pressure behind the hydraulic oil pump
The next thing is, we want to evaluate the effect of the pressure increase
behind the pump. As the level in the upper tank in Fig. 5.2 continues to
increase, the pressure in the line between the tanks also increases. We have
to apply an increasingly higher pressure with the pump to move the fuel oil
to the second tank. However, we have specified in our model in Fig. 5.3 that
the speed of the pump should be kept constant. Therefore, an ever-
increasing torque must occur at the pump shaft during the pumping process.
This expected result is also shown in the graph in Fig. 5.3.
Problems
Let’s look at some problems we can explore with the digital twin of the two
connected tanks.
5.1 What happens to the simulation when the bottom tank is empty?
Investigate how long you can run the simulation. Discuss the result.
5.2 How do the results change when you increase the footprint of the
lower tank from 0.5 [m] to twice the value of 1.0 [m]?
5.3 Operate the experimental setup with water instead of fuel oil. What do
you have to change? Compare the results.
5.4 Investigate what happens when the speed of the pump is doubled.
5.5 Try to build a network of tanks with a slope of the pipes between
them.
Important: For liquid lines, the slope is very important because the
liquid can only flow downwards if there is no pressure difference. If the
liquid has to flow upwards, we need additional pressure, e.g. from a
pump. Therefore, when creating the line, you must pay attention where
the starting point (1) and the end point (2) are located. Specify the angle
of the line between these starting and end points. Negative angles are
also possible or the pipes can also be connected in reverse.
Item Parameter
STEP0 Value after step = 100 [zero]
Step time = 1 [s]
FP04 Index of hydraulic fluid = 0
Temperature = 40 [degC]
Density = 850 [kg/m**3]
Bulk modulus = 17000 [bar]
Absolute viscosity = 51 [cP]
HJ000 Pressure at port 1 = 40 [bar]
Pressure at port 2 = 0 [bar]
Rod velocity = 0 [m/s]
Rod displacement = 0.3 [m]
Index of hydraulic fluid = 0
Piston diameter = 25 [mm]
Rod diameter = 12 [mm]
Item Parameter
Length of stroke = 0.3 [m]
Dead volume at port 1 end = 50 [cm**3]
Dead volume at port 2 end = 50 [cm**3]
Total mass being moved = 250 [kg]
Angle rod makes with horizontal = 90 [degree]
Coulomb friction force = 0 [N]
Stiction force = 0 [N]
Viscous friction coefficient = 0 [N/(m/s)]
Leakage coefficient = 0 [L/min/bar]
HYDROF0 Index of hydraulic fluid = 0
Number of parallel orifices = 1 []
Orifice geometry = circular []
Diameter = 3 [mm]
HL0002 Index of hydraulic fluid = 0
Section type = circular
Diameter = 25 [mm]
Length = 1 [m]
Relative roughness = 1e-05 []
HA001 Pressure at port 1 = 40 [bar]
Index of hydraulic fluid = 0 []
Adiabatic initialization = 1 []
Gas precharge pressure = 40 [bar]
Accumulator volume = 0.5 [L]
Polytropic index = 1.4 [zero]
TK000 Tank pressure = 0 [bar]
In order to improve the jack’s function, an additional restriction should
be placed between the pressure vessel and the hydraulic piston to limit the
flow and reduce pressure pulsations. Set the simulation time back to 10 [s]
according to Table 5.4.
Table 5.4 Simulation time for the simulation model of a hydraulic jack
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.01 [s]
5.3.3 Results and Analysis
The simulation runs for exactly 10 [s] according to the specification given
in Table 5.4. After the first second, the jack is pushed down from the top
with a force of 100 [N]. As can be seen in Fig. 5.5, this happens suddenly.
The signal occurs within a very short time window. Due to its viscosity, the
hydraulic oil cannot react as quickly as possible. This is noticeable in an
oscillation of the oil pressure.
Fig. 5.5 Force on the jack—after one second the force is increased from 0 to 100 [N]
Fig. 5.6 Downwards movement of the hydraulic piston in the jack
Fig. 5.7 Pressure oscillation—pressure in pressure accumulator increases from 50 to 52 [bar]
We see in Fig. 5.6 that the piston does move downwards in accordance
with the force signal. However, it responds with an oscillation at the end of
the force application due to the damping of the oil volume inside the piston.
The oscillation continues through the pipeline. The components “phone”
each other.
In the accumulator tank, we can see an increase in pressure from 50 to
over 52 [bar]. When the force decreases, the piston can then release the
pressure from the accumulator tank. However, in Fig. 5.7, we also see a
clear pressure oscillation in the accumulator despite the throttle. We can
now perform some experiments with the digital twin of the jack in Fig. 5.4.
We can investigate the following work suggestions for this hydraulic model
on our own.
Problems
What can we investigate with our model if we want to lift up a car? Have a
look at the following problems.
5.6 How do the results change when the moving mass increases or
decreases? Use masses with 150, 250, and 500 [kg].
5.8 How does the pressure oscillation change if you increase or decrease
the throttle cross section?
5.9 Change the material values of the hydraulic oil. What would happen if
you ran the jack with water? Does that make sense technically?
References
1. Idel’cik, I.E.: Handbook of Hydraulic Resistance, 3rd edn. Begell House Inc., (1996)
2.
Miller, D.S.: Internal Flow Systems, 2nd edn. Amazon Technology (1989)
3.
Bode, B.: Verfahren zur Extrapolation wichtiger Stoffeigenschaften von Flüssigkeiten unter
hohem Druck. Tribol. Schmierungstech. 37(4), 197–202 (1990)
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_6
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
The word pneumatic comes from the ancient Greek pneuma and means
breath or wind. In creating the digital twin, we examine how gases can be
transported inside pipe systems. When using the pneumatic library, you
should keep in mind that gases are usually compressible. This can lead to
higher computational effort and a worse convergence of the mathematical
equation system than in incompressible problems. It is also difficult to
calculate complicated phenomena such as vortexes or turbulence. To model
turbulence, one should use another simulation tool such as ANSYS CFX or
ANSYS Fluent [1].
Item Parameter
PNGD00 Temperature at port 1 = 293.15 [K]
Fluid definition = air
PNCS001 Index of hydraulic fluid = 0 []
Pressure at port 1 = 1 [barA]
PNGD00 Gas type index = 2 []
Fluid definition = methane (CH4)
PNCS001 Temperature at port 1 = 293.15 [K]
Pressure at port 1 = 3 [barA]
PNCV001 Gas type index = 1 []
Orifice area = 5 [mm**2]
Check valve cracking pressure = 1.8 [bar]
Hysteresis for opening/closing = 0 [bar]
PNCH022 Temperature at port 1 = 293.15 [K]
Pressure at port 1 = 1.013 [barA]
Gas type index = 2 []
Volume = 10 [L]
Thermal exchange coefficient = 500 [J/m**2/K/s]
Thermal exchange area = 0.1 [m**2]
Exchange temperature = 293.15 [K]
PNL000R Gas type index = 2 []
Diameter of pipe = 10 [mm]
Pipe length = 10 [m]
Relative roughness = 1e-05 [zero]
Table 6.2 Simulation time for the simulation model biogas tank and safety valve
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.1 [s]
Fig. 6.2 Pressure in the biogas tank and mass flow of the gas through the safety valve
6.1 Lower the pressure of the methane gas at the inlet. Observe what
happens.
6.2 Increase and decrease the volume of the tank and observe how this
affects the pressure curve in the tank.
6.3 Change the opening behavior of the safety valve by decreasing the
pressure at which the valve opens. What happens to the pressure in the
tank?
6.4 How do the results change if you use propane instead of methane as
biogas? What changes if you simulate with air?
Table 6.3 Parameters for the simulation model of a ventilation system (part 1)
Item Parameter
PNGD00 Gas type index = 1 []
Fluid definition = air
PNCS001 Temperature at port 1 = 293.15 [K]
Pressure at port 1 = 1.013 [barA]
PNCP00 Gas type index = 1 []
Item Parameter
Compressor displacement = 1000 [cc/rev]
Polytropic constant = 1.35 [null]
PNPC1 Gas type index = 1 []
Diameter of pipe = 8 [cm] or 15 [cm]
Pipe length = 5 [m]
Relative roughness = 1e-05 [null]
PNBP001 Gas type index = 1 []
Hydraulic diameter = 8 [cm]
Curvature radius = 10 [cm]
Center angle = 60 [degree]
Relative roughness = 1e-05 [zero]
Item Parameter
TPTE001 Gas type index = 1 []
Diameter at port 1 = 15 [cm]
Diameter at ports 2 and 3 = 20 [cm]
Friction factor in the main branch = 0.1 [null]
Friction factor side branch = 1.2 [null]
PN3P000 Gas type index = 1 []
Pressure drop coefficients = Idelchik
Side branch diameter (port 1) = 15 [cm]
Straight passage diameter (port 2 and 3) = 15 [cm]
Angle between side branch and straight passage = 90 [degree]
Critical Reynolds number = 5000 [zero]
Time constant = 1e-06 [s]
Transition accuracy = 0.9 [null]
PNL0000 Gas type index = 1 []
Model = polytropic
Diameter of pipe = 10 [cm], 15 [cm] and 20 [cm]
Pipe length = 1 [m], 2 [m], 3 [m] and 20 [m]
Polytropic constant = 1.35 [zero]
Table 6.5 Simulation time for the simulation model of a ventilation system
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.1 [s]
Fig. 6.4 Mass flow rate of the air at the compressor at different speeds
In Fig. 6.4, we look at the mass flows of air at the end of the three
branches of the ventilation system. There is a different mass flow in each
pipe. Depending on the length and diameter of the pipes, but also due to
diversions, there are different pressure drops. This further complicates the
behavior of the pipe system.
Fig. 6.5 Mass flow of the air at all three outlets of the ventilation system
You can see the air distribution at the three outlets of the duct system in
Fig. 6.5. Interpret what this means for the ventilation system. Is it desirable
to have a strong uneven distribution of air flows? How does this affect the
indoor climate inside the building? Also work on the following suggestions.
Problems
We want to look at following problems and solve them with the model of
the ventilation system:
6.5 Try changing the speed distribution on the compressor. How does this
affect the airflow?
6.6 Does the air distribution stay the same if they keep increasing the
speed of the compressor?
6.7 Investigate how the air distribution changes when they make
individual pipes thinner.
6.8 Each of the ventilation pipes should have a slider. You can open and
close it. Sliders result in changed pressure drops at the outlets, on which in
turn the volume flow depends. How can they create a uniform air
distribution?
6.9 Create additional flow channels and try to adjust them so that the air is
distributed as evenly as possible at the outlets.
Reference
1. https://www.ansys.com/
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_7
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Simulation settings
Start time = 0 [s]
Final time = 9 [s]
Print interval = 0.001 [s]
The simulation times for the twin of the permanent electrical motor with
load are given in Table 7.2.
Fig. 7.3 The torque [rev/min] and speed [Nm] of the electric permanent DC motor
On one hand, the relative motor speed due to the different currents is
shown. After each step, an oscillation of the rotor speed can be seen, even if
the electromagnetic torque remains more or less constant. It reacts to the
oscillations in the generator’s magnetic field.
Problems
For our model of the electric motor we can make several working
suggestions. Take a look at a few work suggestions that we can explore with
our twin: To do this, they should find and modify the appropriate model
parameters in the simulation model.
7.1 Change the armature winding resistance to 1.0 [Ohm] and to 2.4
[Ohm]. Explain what happens to the results.
Fig. 7.4 Asynchronous squirrel-cage induction machine with three phase and mechanical load
Table 7.4 Simulation time for the simulation model of the electric three-phase motor
Simulation settings
Start time = 0 [s]
Final time = 0.04 [s]
Print interval = 0.0001 [s]
Fig. 7.5 Phase-shifted voltage curves of the asynchronous motor with three phase and mechanical
load
The three-phase electrical ports are associated on the icon with the
notations I, II, and III. The star and delta connections for the windings A, B,
and C as well as the winding voltages and currents are defined within this
convention.
Problems
Let’s look at some problems that we can now investigate with our electric
motor twin. To do this, you should modify the corresponding model
parameters in the simulation model.
7.5 Change frequency of the three sinusoidal signal functions up to 100
[Hz].
7.7 Change the rotor time constant to 0.165 [s]. What happens?
7.8 Change the load of the model and change the limit temperature of the
armature of the motor. Give an interpretation of your results.
Fig. 7.7 Checking the boundary conditions of the generator’s torque based on mechanical drive
Fig. 7.8 Changes of the current and voltage steps based of the production of the generator
Fig. 7.9 Joule’s losses produce a heat source in the housing of the generator
The temperature change in the housing is influenced by the mass of the
housing [THC000] and the material. Here in Fig. 7.9 we used aluminium,
but alternative materials like iron or cats iron can also be used. We can see
how the temperature changes inside the generator housing.
Cooling of housing for electric generators as well as the construction
and design of fins for heat conductors is an important topic. It is widely
used for different applications like wind turbines, power plant, and other
fluid energy machines. Here we see a good example of the connection
between the thermal library and the electric library.
Problems
We want to have a look at some problems that we now can explore our
generator. To do this, you should modify the simulation model.
7.9 Change the thermal mass that is connected to the housing to 5 [kg]
and to 10 [kg] and compare the results. Use cast iron as alternative housing
material. Discuss the results.
7.12 Try to build a convective cooling unit with the thermal library to
cool down the electric housing.
Reference
1. IEC 60050 Rotation Machinery – General, IEV ref. 411-31-10: Induction Machine – an
asynchronous machine of which only one winding is energized. Publication date: 1990-10.
Section 411-31
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_8
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Fig. 8.2 Digital twin for liquid piston compressor with one cylinder for multiple strokes
The role of the relief valve [RV010] is to limit the upstream pressure
within a hydraulic circuit and thus protect hydraulic components from over
pressure. This component is also known as pressure limiting valve,
maximum-pressure valve, or safety valve [5].
Item Parameter
PNCH022 Temperature at port 1 = 293.15 [K]
Pressure at port 1 = 1.013 [barA]
Gas type index = 1
Volume = 0.1 [L]
Thermal exchange coefficient = 0 [J/m**2/K/s]
Thermal exchange area = 0.1 [m**2]
External temperature = 293.15 [K]
PNCV010 Gas type index = 1
Check valve cracking pressure = 0.2 [bar]
Check valve mass flow rate pressure gradient = 10 [g/s/bar]
Valve hysteresis = 0 [bar]
PNGD00 Gas type index = 1
Fluid definition = air
Properties definition = semi perfect
PNAS001 Temperature at port 1 = 293.15 [K]
Item Parameter
PNJ002 Temperature at port 1 = 293.15 [K]
Pressure at port 1 = 1.013 [barA]
Gas type index = 1
Model = with thermal exchange
Use initial displacement = yes []
Displacement of piston = 0.5 [m]
Piston diameter = 25 [mm]
Rod diameter = 12 [mm]
Length of stroke = 0.5 [m]
Dead volume at port 1 end = 50 [cm**3]
Thermal exchange coefficient = 0 [J/m**2/K/s]
External temperature = 293.15 [K]
Viscous friction coefficient = 0 [N/(m/s)]
Leakage coefficient = 0 [g/s/barA]
Spring preload = 0 [N]
Spring rate = 1 [N/m]
Pressure in rod chamber = 1.013 [barA]
Spring rate at endstops = 100000 [N/mm]
Damping coefficient at endstops = 100000 [N/(m/s)]
Deformation on endstops at which damping rate = 0.001 [mm]
Table 8.2 Parameters for simulation model of the liquid piston compressor with one cylinder for one
stroke (part 2)
Item Parameter
RV010 Index of hydraulic fluid = 0
Relief valve cracking pressure = 5 [bar]
Relief valve flow rate pressure gradient = 500 [L/min/bar]
Valve hysteresis = 0 [bar]
Item Parameter
HJ0023 Pressure at port 1 = 1.013 [bar]
Index of hydraulic fluid = 0
Use initial displacement = yes
Displacement of piston = 0 [m]
Piston diameter = 25 [mm]
Rod diameter = 12 [mm]
Length of stroke = 0.5 [m]
Dead volume at port 1 end = 50 [cm**3]
Viscous friction coefficient = 0 [N/(m/s)]
Leakage coefficient = 0 [L/min/bar]
Spring preload = 0 [N]
Spring rate = 1 [N/m]
Pressure in rod chamber = 1 [bar]
Spring rate at endstops = 100000 [N/m]
Damping coefficient on endstops = 10 [N/(m/s)]
Deformation on endstops at which damping rate is fully effective = 0.001 [mm]
PU001 Index of hydraulic fluid = 0
Pump displacement = 2 [cc/rev]
Typical pump speed = 1500 [rev/min]
PM000 Shaft speed = 1500 [rev/min]
FP04 Type of fluid properties = elementary
Index of hydraulic fluid = 0
Temperature = 40 [degC]
Name of fluid = unnamed fluid
Density = 850 [kg/m**3]
Bulk modulus = 17000 [bar]
Absolute viscosity = 51 [cP]
Item Parameter
MECMAS21 Velocity at port 1 = 0 [m/s]
Displacement at port 1 = 0 [m]
Use friction = no
Endstop type = none
Mass = 0.1 [kg]
Inclination (+90 port 1 lowest, −90 port 1 highest) = 0 [degree]
TK000 Tank pressure = 0 [bar]
The existing model must now be parameterized accordingly. In the
following only those parameters are shown, which deviate from the basic
settings. The simulation times for the simulation model are given in Table
8.3.
Table 8.3 Simulation time for the simulation model of the liquid piston compressor
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.01 [s]
Item Parameter
MECDSS0B Sign convention: positive from = 0
Offset to be subtracted from displacement = 0 [m]
Gain for signal output = 1 [1/m]
TRIG0 Initial output value = low
High input threshold value = 0.499 null
Low input threshold value = 0.001 null
High output value = 0 null
Low output value = 1 null
HSV23 02 Index of hydraulic fluid = 0
Valve rated current = 1 [mA]
Valve natural frequency = 80 [Hz]
Valve damping ratio = 0.8 null
Ports P to A characteristic flow rate at maximum opening = 100 [L/min]
Ports P to A corresponding pressure drop = 0.01 [bar]
Ports P to A critical flow number (laminar/turbulent) = 100 null
Ports A to T flow rate at maximum opening = 100 [L/min]
Ports A to T corresponding pressure drop = 0.01 [bar]
Ports A to T critical flow number (laminar/turbulent) = 1000 null
Time steps and simulation time for the liquid piston compressor with
multiple stroke are given in Table 8.5.
Table 8.5 Simulation time for the liquid piston compressor with multiple strokes
Simulation settings
Start time = 0 [s]
Final time = 10 [s]
Print interval = 0.01 [s]
Fig. 8.7 Displacement of the mass element between hydraulic and pneumatic pistons with multiple
strokes
In Fig. 8.9, the pressure in the pneumatic and hydraulic cylinder is
shown. The hydraulic cylinder after the first stroke retracts again due to the
residual pressure in the pneumatic cylinder and the spring force. The
pneumatic cylinder draws in new air.
Fig. 8.8 Digital twin for liquid piston compressor with two cylinder and multiple strokes
When the hydraulic cylinder is almost completely retracted, the process
is repeated. With each conveying process, the pressure in the pressure
accumulator increases step by step. It can be seen that pressure fluctuations
inside the hydraulic fluid are rather high. In technical applications, this
behavior should normally be avoided (see Fig. 8.10).
Fig. 8.9 Pressure curves in hydraulic, pneumatic cylinder with multiple stroke and high oscillations
Fig. 8.10 Pressure curves in the pressure accumulator chamber with multiple strokes
Now, we added a two-position four-port hydraulic valve [HSV24 02]
to the model to switch between the two different hydraulic cylinders. This
valve is controlled also by the trigger signal [TRIG0].
Figure 8.12 shows the pressure curves of the pneumatic actuator on the
left and the pressure curve of the pneumatic actuator on the right. It should
be mentioned that we have not visualized the pressure variations in the
hydraulic chamber as in the previous section, but they are comparable. The
pressure in the chamber of the pressure accumulator is shown in Fig. 8.13.
It can be seen that there are many more stages forming the stepwise
pressure rise in the pressure accumulator chamber.
Fig. 8.13 Pressure curve in the pressure accumulator chamber with multiple strokes
We want to analyze our complex technical system and have a look at
some problems to improve the model. You can modify the appropriate
parameters in the model and try out what’s happening.
Problems
8.1 Change the size of the pressure accumulator from 0.1 to 10 [L]. What
happens?
8.2 Change the trigger signal of the sensor and try to reduce the
fluctuations inside the hydraulic chamber.
8.3 Add an additional third hydraulic and pneumatic actuator and repeat
the simulations.
8.4 Examine the temperature of the gas inside the pneumatic accumulator.
Is it adiabatic? How can you change the model to reduce the temperature
inside the accumulator?
8.5 Think about a new component to store the heat energy of the gas in
the accumulator.
8.6 Add a hydraulic chamber in the back flow area between the 2/4-way
hydraulic valve and the pump.
References
1. Staudacher, T., von Roon, S., Vogler, G.: Energy storage - status, perspectives and economic
viability; study summary report; client boarfeleven; performed by Forschungsstelle für
Energiewirtschaft e.V., 03.2009
2.
German Federal Ministry for Economic Affairs and Energy (The energy of the future, 2021): 8th
Monitoring Report on the Energy Transition – Reporting years 2018 and 2019, Access: 2021-05-
09. https://www.bmwi.de/Redaktion/DE/Publikationen/Energie/achter-monitoring-bericht-
energie-der-zukunft.html
3.
Grün, S., Hübner, D.H., Ortwig, H.H., Rückert, F.U.: Investigation and optimization of a three
stage inter-cooled piston compressor for an energy storage system with pressurized air. In:
Proceedings of the Interdisciplinary Conference on Mechanics, Computers and Electrics
(ICMECE 2022), 6–7 October 2022, Barcelona, Spain
4.
Grün, S., Hübner, D.H., D., Molter, J.: Investigation and simulation based optimization of an
energy storage system with pressurized air. Arch. Thermodyn. 42(4), 183–200 (2021). https://doi.
org/10.24425/ather.2021.139658. https://journals.pan.pl/Content/122213/PDF/art11_corr.pdf
5.
Siemens Digital Industries Software: Simcenter Amesim Student Edition (2020)
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_9
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Fig. 9.4 Photoreceptors of a frog are now connected to three neurons that initiate an action when
they are sufficiently excited [2]
There are three events that are important for the frog. The frog sees a
stork, a fly, or just the blue sky. Each time, the photoreceptors are excited in
a characteristic way. The photoreceptors are now connected to three
neurons that initiate an action when they are sufficiently excited.
Fig. 9.5 Definition file frog example.net for the frog with training and test data
Fig. 9.6 Enter the input data for the experience of the frog’s life and hunting behavior
We now graphically represent the connection between the three visual
cells and the three nerve cells, the connecting nerve fibers and the synapses.
In each of the synapses, their strength is plotted as a value between -1 and
1. Each of the three nerve cells fires when the sum of the transmitters
arriving via the synapses exceeds the measure 0.8 in an interval from 0 to 1
(see Fig. 9.6).
The model [DYNNNFF01] can be compared with the [DYNPSM02].
For now, this process needs to be done using the MathWorks Neural
Network toolbox for MATLAB [5]. It is also a submodel that allows creating
global approximations, and it is especially suitable in presence of non-
smooth or highly non-linear models. The input file listed in Table 9.1 has to
be generated by MATLAB or by our in-house programmed optiflow neural
network tool.
The only drawback of this submodel is that it currently has a
dependency to the commercial code MATLAB which is rather slow for
creating the artificial neural network, but not for evaluating it. Note that,
just as [DYNPSM02], [DYNNNFF01] only uses standard Amesim XY
tables, which is particularly interesting to create a surrogate model that can
be efficiently exported as a Simulink blackbox, as a Functional Mock-up
Unit (FMU), or executed on a real-time target. Examples for the XY-tables
needed are given in Table 9.1.
Table 9.1 Data files written by optiflow neural network tool for Simcenter Amesim
Port Description
Port 1 Used to supply inputs to the neural network model
Port 2 Provides the responses estimated by the neural network model
Fig. 9.7 Software tool optiflow neural network (ONN) can be used to create weights
Fig. 9.8 Testing your neural network of the frog’s example
Under the topic Connectcalls you define how to connect the sensor cells
and neurons against each other. The line (1 3 4 13) means that all three
input cells have one connection to the hidden layer (4 to 13). And all hidden
neurons (4 to 13) have a connection to the output cells (14 to 16). After
loading the example file frog example.net you can start the training of
your frog brain.
Problems
We want to have a look at some problems that we now can explore with an
artificial neural network. To do this, you try the following things with our
froggy.
9.1 Create a data set for the learning behavior of the frog. It should
contain 100 entries of training data. Verify the results with your own test
data.
9.2 Train your neural network with the frog information data set. And try
to predict the behavior.
9.3 Visualize the test cases in a diagram and compare input and output
data. Discuss the result.
9.4 Try to create a mesh with 20 hidden layers and compare the results of
both meshes.
References
1. Spitzer, M.: Lernen – Gehirnforschung und die Schule des Lebens. Elsevier (2007)
2.
Mathematik für Wirtschaftsingenieure; Skript, Hochschule für Wirtschaft und Technik des
Saarlandes (htw saar) (2020)
3.
LMS Imagine.Lab Amesim Users Guide, Siemens Industry Software S.A.S (2016)
4.
https://www.optiflow.htwsaar.de/
5.
https://www.mathworks.com/help/matlab/ref/rand.html
OceanofPDF.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
F. U. Rückert et al., Digital Twin Development
https://doi.org/10.1007/978-3-031-25692-9_10
10. Conclusions
Frank U. Rückert1 , Michael Sauer2 , Tuomo Liimatainen3 and
Dirk Hübner4
(1) Fluid Energy Machines, University of Applied Sciences Saarbrücken,
Saarbrücken, Saarland, Germany
(2) Fluid Machinery and Measurement, University of Applied Sciences
Saarbrücken, Saarbrücken, Saarland, Germany
(3) Mechanical Engineering, LAB University of Applied Sciences,
Lappeenranta, Etelä-Suomi, Finland
(4) Lightweight Construction, University of Applied Sciences
Saarbrücken, Saarbrücken, Saarland, Germany
Michael Sauer
Email: [email protected]
Tuomo Liimatainen
Email: [email protected]
Dirk Hübner
Email: [email protected]
Then the teams create a technical scenario with the digital twins and
changed the parameters and constraints to produce different results for the
same model. Based on the results, roles and rules need to be clearly defined
in two scripts. As with the hunting scene from the petroglyphs in Bryce
Canyon, the moderator with the antlers handed the two alternative scripts
with the different results to the two teams and gave them the task of
discussing them and trying to convince the other side. Teams could act out a
specific discussion in this safe environment.
Our idea was that this could be a dynamic learning experience for actors
to gain knowledge about physical behavior by immersing them in simulated
real-world problems. But also in industry, such an artificial dispute in a
safe, moderated environment can help teams learn very quickly and focus
on a technical problem. Adrenalin is provided free of charge and no
additional coffee is needed.
References
1. https://www.ua.pt/en/news/9/69847
2.
Russell, C., Shepherd, J.: Online role-play environments for higher education. Br. J. Educ.
Technol., Special Issue: Learning Objects in Progress 41, 6 (2010)
3.
https://www.plm.automation.siemens.com/global/en/
OceanofPDF.com