Conference PaperPDF Available

Modeling techniques for integrated simulation of industrial systems based on hybrid PDEVS

Authors:

Abstract and Figures

The push for energy efficiency of industry processes is driving various efforts to analyze, simulate and optimize the underlying complex and large cyber-physical systems. While some efforts use co-simulation, we instead focus on an integrated hybrid approach that offers the ability to model hybrid components as a whole with all their aspects, based on Hybrid PDEVS. We describe modeling techniques that were developed for this purpose, and demonstrate the feasibility of the approach with a prototypical example. The proposed approach allows describing hybrid models on component level with improved reusability.
Content may be subject to copyright.
Modeling Techniques for Integrated Simulation of
Industrial Systems Based on Hybrid PDEVS
Philipp Raich, Bernhard Heinzl, Franz Preyser, Wolfgang Kastner
Institute of Computer Aided Automation
Automation Systems Group
TU Wien, Vienna
{praich, bheinzl, fpreyser, k}@auto.tuwien.ac.at
Abstract—The push for energy efficiency of industry processes
is driving various efforts to analyze, simulate and optimize the
underlying complex and large cyber-physical systems. While
some efforts use co-simulation, we instead focus on an integrated
hybrid approach that offers the ability to model hybrid compo-
nents as a whole with all their aspects, based on Hybrid PDEVS.
We describe modeling techniques that were developed for this
purpose, and demonstrate the feasibility of the approach with a
prototypical example. The proposed approach allows describing
hybrid models on component level with improved reusability.
Index Terms—hybrid modeling, energy efficiency, discrete-
event simulation, integrated simulation, Hybrid Parallel DEVS,
production process, industrial systems, Balanced Manufacturing,
BaMa, Cube
I. INTRODUCTION
Energy and resource efficiency is becoming increasingly
important in the industrial sector because of its economical
and ecological impact and at the same time significant poten-
tial for savings. Several approaches to this problem employ
simulation-based techniques to tackle the complexity of these
systems [1, 2], yet covering such systems as a whole, incorpo-
rating aspects from different domains (production machinery,
energy infrastructure, building, etc.), has proven difficult, as it
requires investigating continuous as well as discrete aspects,
resulting in interdisciplinary cyber-physical systems.
In the ongoing Balanced Manufacturing (BaMa) project1,
we aim at improving energy costs and footprint of manufac-
turing processes through monitoring, simulation and optimiza-
tion [3]. The goal is to seamlessly include simulation function-
ality as part of the BaMa tool-chain into existing automation
systems without relying on proprietary software or libraries.
Since usability in terms of modeling effort, maintainability and
flexibility is of great importance in order to enable widespread
application, we decided to opt for an integrated modeling and
simulation approach, in contrast to popular approaches using
co-simulation.
Representing the underlying system as a whole, with con-
tinuous and discrete aspects in one simulation environment,
requires one single hybrid formalism and further allows de-
scribing self-contained, loosely coupled and reusable hybrid
model components. The formalism also has to be open and
1http://bama.ift.tuwien.ac.at
standardized to allow a transparent implementation of the
simulation engine.
After evaluating different formalisms and descriptions [4],
Hybrid Parallel DEVS (Hybrid PDEVS) [5] as a hybrid
formalism based on Parallel DEVS (PDEVS) [6] was chosen.
We present our findings and the modeling techniques we
developed by using this approach and further demonstrate
an example implementation of a production facility. This
paper is organized as follows: Related work is presented in
Section II and an introduction to the used formalisms is given
in Section III. Section IV describes the modeling process and
applied techniques, which are demonstrated with an example
in Section V. Our findings are discussed in Section VI and
concluded in Section VII.
II. RE LATE D WORK
As already mentioned, some approaches for simulating
hybrid systems employ co-simulation by coupling different
simulation environments via middleware [7, 8]. For example
the High-level Architecture (HLA) [9] specifies a standard for
distributed simulation including interface and model template
specifications. Apart from the computational overhead, this
type of simulation usually introduces significant complexity
into the modeling process, in particular regarding model
development, maintenance and overall usability of existing
components. In many cases, these types of co-simulations
are highly customized to a particular application (e.g. smart
grids [10]) and/or simulation environments.
Efforts in this area from the Modelica community try to
establish the Functional Mock-up Interface (FMI) [11] as
a common standard for co-simulation and model exchange.
FMI became quite powerful in recent years and has gained
popularity across different simulation tools, it is however
focused predominantly on continuous models based on the
Modelica language. Employing Modelica and FMI for our
research would have entailed a few drawbacks: (1) While
Modelica allows to also incorporate hybrid characteristics in
terms of state events, this was not sufficient for our attempts
to model discrete persistent entities in a native and seamlessly
integrated way. (2) Regarding simulation algorithms, Modelica
allows less freedom for combining ODE solvers and discrete-
event schedulers, resulting in reduced performance.
Other research efforts are attempting to combine process-
oriented modeling with Modelica, resulting for example in the
Modelica DESLib library2, parts of which are also based on
PDEVS. Evaluation of the library in an attempt to adopt it for
our own research however showed significant downsides (e.g.
usability, entities).
Regarding research related to DEVS, Goldstein et al. [12]
introduce modifications to the Discrete Event System Speci-
fication (DEVS) to make the formalism more appealing and
convenient to prospective users. In contrast, we aim at leaving
the formalism untouched and present modeling considerations
based on top of the formalism.
III. HYBRID PDEVS FOR MA LI SM
Discrete Event System Specification (DEVS) is a mathe-
matical formalism for modeling and analysis of discrete event
systems that was introduced by Zeigler [13]. DEVS distin-
guishes between atomic and coupled components to compose
models in a modular and hierarchical manner. While atomic
models are the prime building blocks (i.e. the leaves in a tree
of hierarchical components) and define the dynamic behaviour,
coupled models describe a set of interacting components,
which can be atomic DEVS or again other coupled DEVS.
To improve handling concurrent events on atomic level,
the classic DEVS formalism was extended to Parallel DEVS
(PDEVS) [6].
Aiming at hybrid systems, Discrete Event and Differential
Equation System Specification (DEV&DESS) [14] (based on
DEVS) as well as Hybrid PDEVS [5] (based on PDEVS) are
extended formalisms incorporating continuous model aspects
into the model description. Comprehensive evaluations [4]
showed Hybrid PDEVS being better suited for our intended
applications.
An atomic Hybrid PDEVS model is defined as follows:
Ahp = (X, Y, S, f, cse , λc, δstate, δint, δext, δconf , λd, ta),
with the sets of inputs X, outputs Yand states S(all of them
may contain discrete as well as continuous values), the rate of
change function fdefining continuous dynamics, the continu-
ous output function λc, the state event condition function cse
for localizing state events, and the six characteristic functions
defining discrete dynamics:
internal transition function δint (defines internal events),
external transition function δext (reactions to external
input events),
confluent function δconf (resolving simultaneous events),
state event transition function δstate
discrete output function λd,
time advance function ta.
In addition, Hybrid PDEVS defines coupled models as:
NP= (XN, YN, D, {Md|dD}, EIC, EOC, IC ),
with sets of set of input and output events XNand YN, set of
subcomponents Mdwith corresponding index set Dand three
2https://github.com/modelica-3rdparty/DESLib
distinct sets describing connections between subcomponents:
EI C for external input couplings, EOC for external output
couplings, and IC for internal couplings. For a more detailed
explanation we refer to [5, 13].
In addition to the model description, the formalisms also
specify a simulation algorithm for executing such models.
An abstract simulation engine for Hybrid PDEVS models has
to incorporate handling discrete events as well as numerical
integration algorithms. For the discrete part, this includes
simulator components for executing atomic events and coor-
dinators for routing event messages and invoking simulators
inside couplings.
For numerically integrating the continuous model, there
are several different approaches. The Quantized State Systems
(QSS) method [15, 16] aims at embedding continuous solvers
by quantizing state values, resulting in discrete-event systems
(instead of traditional discrete-time systems).
In contrast, an ODE wrapper concept developed by [5]
generates a closed representation of continuous model parts -
modeled as Ordinary Differential Equations (ODE) and alge-
braic equations - of all atomics at runtime, which can then be
executed by a single ODE solver. The idea is to be able to use
established and sophisticated numerical methods for solving
differential equations. The ODE wrapper function does not
modify the modular hierarchical model itself but rather creates
an additional data structure alongside the discrete simulation
engine. During runtime the continuous cycle is iterated until
the next discrete internal event becomes imminent or a state
event occurs, at which point respective discrete events are
executed until simulation time can advance, again starting with
the continuous integration cycle.
For implementation of the prototype model presented in this
paper (see section Section V), we pursued the ODE wrapper
approach for which we utilized an available framework as part
of the MatlabDEVS Toolbox3[17].
IV. MODELING
A. BaMa Cubes
One core concept of BaMa is the “Cube”, which is used to
divide the overall system into manageable parts to tackle the
complexity of the model. Cubes are used to demarcate energy
and resource boundaries in a system, and thus must implement
the necessary interfaces to communicate with neighboring
cubes, i.e. their siblings and parents/children. The communica-
tion between cubes can be either energy-related, e.g. transfer
of waste energy, but also consist of control messages or
entity/event information. As such, cubes are most often virtual
equivalents of physical subsystems of the real system, e.g.
an oven, machine tool, thermal zone, heater, etc. The cubing
concept is covered more detailed in [3].
B. Example Cube: An Oven
A cube is defined by two distinct facets, its outer structure
and its inner behavior, and we decided to look at both facets
3The MatlabDEVS Toolbox is available at https://www.mb.hs-wismar.de/
cea/DEVS Tbx/MatlabDEVS Tbx.html.
separately during modeling. As an example, we present the
modeling process of a simple conveyor oven.
The outer structure of a cube is composed by its interfaces
over which it communicates with other cubes. Figure 1 shows
the graphical depiction of the interfaces of an oven cube.
Demand electrical powe r (PelD)
Demand thermal power (Q
HD)
Entity in ACK (EINcom )
Oven
Electrical power (Pel)
Thermal power (Q
H)
Entity in (EIN)Entity out (EOUT)
Waste entity (EW)
Waste heat (Q
WH)
Recovered heat (Q
rec)
Capacity (N)
Production schedule (Pplan)
Nominal power (PH)
Standby power (Ps)
Holding period (tB)
Set temperature (Ts)
Hysteresis (H)
Sign heating/cooling (sign)
Volume (V)
Heat transition (UA)
Heat capacity air (cpA)
Air density (ρA)
Waste heat utilization (η)
Waste fraction (α)
Parameters:
State variables:
State (p)
Heating state (h)
Entities (ent)
Temperature (T)
Ambient temperature (Ta)
Entity out ACK (EOUTcom )
Waste entity ACK (EAcom)
Fig. 1. Interfaces of an oven cube showing inputs (left) and outputs (right).
The figure also shows parameters and variables of the internal model.
The inner behavior is usually more complex to model. For
the example at hand, this involves its continuous behavior, e.g.
the energetic behavior of the oven defined by balance equa-
tions and other differential and algebraic equations (c.f. Equa-
tions (1) and (2)) for energy-related internal variables (e.g.
temperature):
dT
dt =˙
QH(TTa)·UA
cpA ·ρA·V+PEent E.cp·E.m (1)
˙
QW H = ((TTa)·U A +Pel)·(1 η)(2)
where Tdenotes the temperature inside the oven, Tathe
ambient temperature, ˙
QHthe heating power input, ˙
QW H
waste heat output, cpA the specific heat capacity for air, ρA
the air density, Vthe air volume inside the oven, Pel the
electric power input, ηthe heat recovery factor, and U A
denotes the heat transition trough the oven walls. The term
PEent E.cp·E.m gives the sum of the heat capacities of all
entities Eent inside the oven.
This description allows incorporating transient dynamic
behavior, which is crucial when analyzing time-dependent
interactions between different cubes and different domains.
In addition, the oven also incorporates discrete behavior
governing the material flow, discrete states and events, in order
to simulate persistent and traceable products and goods (e.g.
workpieces). This behavior can be described semi-formally
using e.g. state diagrams, see Figure 2.
C. Modeling Considerations
Based on the Hybrid PDEVS formalism and (informal)
cube descriptions, we applied some modeling techniques and
defined considerations, in order to successfully implement a
larger model. Please note that these considerations thus apply
to models based on Hybrid PDEVS and, for most parts,
DEVS/PDEVS, but not necessarily for other discrete event
simulation formalisms or engines.
sta ndby
hol ding o utpu t
EOUT = ent(N); EA = ent(N)
EOUT.T=T; EOUT.m=(1-a)*ent(N).m
EA.T=T; EA.m=alpha*ent(N).m
off
Initial
inco ming
ent(1) = EIN
EINcom = TRUE
upd ate
ent(i+1) = ent(i)
heat ing
wait ing
[ent(N) != 0]
[ent == empty]
[t >=tB/N]
[Pplan signal]
[Pplan signal]
[TRUE]
[EIN & ent(1) == 0]
[EOUTco m & EAcom]
[Pplan signal]
[Pplan signal]
[EIN & ent(1) == 0]
[ent(N) == 0]
[TRUE]
[Pplan signal]
[Pplan signal] [sign*(T-Ts)>=0]
[t >= tB/N]
Fig. 2. State diagram describing the discrete behavior of the oven cube
1) Push vs. Pull: The way entities can be exchanged
between cubes can be divided into two core principles, push
and pull. While both imply the use of a control path be-
tween sender and recipient (see Figure 3), they have different
implications on which precautions must be taken to ensure
correctness.
Entity
Entity
Oven
ACK / REQ
Cube
Entity
Fig. 3. Push/pull principle for exchanging entities between cubes
The push principle describes the method where the preced-
ing holder of the entity will send the entity along its path as
soon as possible, regardless of the successor’s state, i.e. actual
demand for the entity. The recipient must then acknowledge
the reception of an entity (ACK, see Section IV-C2).
The pull principle reverses the aforementioned flow, so that
the successor becomes a requester for the entity (REQ). Al-
though this averts precautions necessary for the push principle,
as cubes will not request entities if they cannot hold them, but
if used excessively, it reverses the control flow through the
whole model.
In the tradition of common material flow simulation tools
(e.g. Plant Simulation4), and due to some severe issues with
the pull principle (see Section IV-C3), we we decided on the
push principle as the primary method for our modeling efforts,
while some aspects of the material flow are more adequately
modeled with pull (e.g. fetching entities from a storage, forklift
behavior, etc).
2) Entity Acknowledgment: Due to the general but rudi-
mentary characteristics of Hybrid PDEVS and its signal prop-
agation, it is necessary to manually ensure that the receiving
block is in a state that would allow him to accept a message.
Hand in hand with the push principle, the simplest reso-
lution is to use explicit acknowledgments for the reception
4http://www.plm.automation.siemens.com/en us/products/tecnomatix/
manufacturing-simulation/material- flow/plant-simulation.shtml
of entities. Accordingly, a sender must retain the entity to
be sent until the acknowledgment arrives. This approach
facilitates material flows where multiple senders concurrently
send entities to the same recipient, eliminating the need for a
superordinate coordinator. It also implies that a sender must
retry sending entities as long as they are not accepted by the
successor, or the entity will remain indefinitely at its place.
We considered the control messages that are necessary
to exchange entities, e.g. acknowledgments, to be logical,
non-physical messages. Thus, they are issued by their origin
instantly, i.e. in the same simulation step the respective entity
was received. Note that in Hybrid PDEVS this causes an
additional iteration at the same simulation time to be initiated.
This is also known as mealy behavior and causes issues of its
own in DEVS [4].
3) Entity Request: The pull principle explained in Sec-
tion IV-C1 reverses the control flow of the explicit acknowl-
edgements of the push principle. The difference lies in the
sequence of the messages and what they imply.
Sending requests for entities means that the prospective
recipients lets its predecessor(s) know that he is now ex-
plicitly awaiting an entity. Considering Hybrid PDEVS as
underlying formalism, this implies that the state of all cubes
“spill” into their predecessors states. This circumvents the
modularity goals of BaMa cubes, creates complications for
Hybrid PDEVS blocks and further causes severe issues at the
initialization of a simulation.
Further, if immediate responses (same time step) are used
for the requests, which we encourage, the REQ output of a
recipient will change during the exchange, i.e. in the same
time instant its state changes immediately from ready to not-
ready, which must again be considered while modeling the
sender. This and issues with multiple senders and/or recipients
make things even more complicated, and again severely limit
the modularity and reusability of the modeled cubes.
4) Concurrent Inputs: Due to the properties of discrete-
event simulation and Hybrid PDEVS in particular (see Sec-
tion III), discrepancies between the behavior of a real cube and
its Hybrid PDEVS description arise while creating a virtual
model. More specifically, concurrent events must be handled
explicitly in order by the virtual representation. These concur-
rent events are especially important considering synchronized
production lines, where external and output messages coincide
and the outputs again cause instantaneous events, including
feedback loops (see Section IV-C2), all handled at different
iterations of the same simulated time step.
Since a Hybrid PDEVS engine will reset all external mes-
sages to a block in-between iterations, even if they happen
at the same simulated time, it is thus necessary to buffer the
external messages for future iterations. The used technique
is to invalidate all buffers as soon as the simulation time
progresses. Else messages might be lost due to concurrent
signals handled first in the input handling routine (δext).
This modeling technique leads to the inputs of a cube being
“implicitly prioritized” and require design decisions to be
made by the modeler. Further, we would argue that corner
cases are easily missed and obscure errors are probable.
V. EX AM PL E
To demonstrate applying the described concepts, we devised
an example of a production plant (shown in Figure 4) using
data and information gathered at a real production facility. The
facility produces baked goods in different variants (fresh as
well as frozen) and includes a production line, energy system
and different thermal zones.
Building
ThermZone2
ThermZone1
ThermZone4ThermZone3
Storage
Mixer
SplitterOven
Freezer
Batcher
Provider
FIFO-Queue
Heater
Cold
Network
Electric
Network
Cooler
ProviderProvider
Heat
Network
Fig. 4. Example production facility consisting of production machines,
logistics components, energy networks and thermal building zones.
A fictitious plant was used instead of a closer representation
of the archetype, because gathering the complete data from a
facility is time-consuming and difficult to justify. The same
applies to the subsequent analysis of the plant. The example
with its reduced complexity does not undermine the main goal
of the exercise, which is to demonstrate the feasibility of the
concept and also demonstrate various critical processes:
Continuous and discrete aspects intertwined
Scheduling of different product types
Distinct product paths
Complex product flows: Splitting, Merging and Batching
A. Model Components
In this example, a building cube is defined, which contains
four thermal zone cubes, each representing a distinct part
of the facility: cold store, offices, plant room and production
hall. The thermal zones all have independent conditioning,
and exchange heat with each other according to the model
parametrization.
The energy system cubes and energy network cubes are
composed of a heater providing heat to oven and thermal
zones, likewise a chiller that supplies cold to the freezer and
thermal zones (both via the respective network) and finally an
electric network.
The production and logistics cubes illustrate a production
line for two product variants. Respective ingredients are pulled
from the storage and sent onto the line, including mixing,
splitting and batching/packaging of entities as well as queues
and conveyor belts. The products either pass an oven for
baking or a freezer for cooling, depending on the type of
product. Specific parameters are applied for different products
to demonstrate process sheet functionality (see below).
The production schedule plays an important role for the
facility model, as it constitutes the major input vector to
differentiate scenarios. Entries in the production schedule are
essentially commands for state changes (see also Figure 2),
and, depending on the cube, different arguments and parame-
ters can be appended. Table I shows two production schedule
scenarios over one day (00:00 to 24:00).
TABLE I
PROD UCT IO N SCH ED ULE S FO R TWO E XA MPL E SCE NAR IO S
Cube Time State Product type Quantity
Scenario 1
Storage 12:00 prepare 1 8
Production 12:30 on 1
24:00 off 1
Oven 11:30 heating 1
23:30 off 1
Scenario 2
Storage 00:30 prepare 1 8
Production 01:00 on 1
07:00 off 1
Oven 00:00 heating 1
07:00 off 1
The process sheet acts as a look-up table for parameters of
a production step for a certain product, e.g. temperature set-
points. In difference to the production schedule, we consider
process sheet parameters to be mutable during simulation run-
time.
Other inputs are also necessary for the model. These could
be data that was acquired from a plant to initialize the model,
or to substitute (outside) effects which are not simulated, e.g.
the temperature outside of the building.
B. Simulation
The aforementioned example was implemented using the
MatlabDEVS Toolbox, based on the semi-formal model de-
scription of the cubes (cf. Section IV-B). Process sheet
and production schedule (see Section V-A) serve as in-
put/simulation parameters and can be adjusted for different
products. The simulation also uses external data to factor in
environment temperatures.
Table II depicts the simulation results from two different
scenarios (see Table I). The left column shows entity count in
production over time (splitting cube resp. oven) and the right
column presents overall energy demand, divided into heating,
cooling and electrical energy.
The scenarios compare the outcome of different production
times of eight product batches. As expected, the results show
different overall energy demands, which is attributed to chang-
ing ambient air temperatures over the course of the day and
also less standby energy consumption due to earlier production
turn off.
TABLE II
SIMULATION RESULTS SHOWING ENERGY COMPARISON BETWEEN TWO
SCENARIOS (SEE TAB LE I)
Production Energy
Scenario 1
0
2
4
6
8
10
splitting.count
0 2 4 6 8 10 12 14 16 18 20 22 24
1
2
3
4
5
time [h]
oven.count
freezer.count
Scenario 2
0
2
4
6
8
10
splitting.count
0 2 4 6 8 10 12 14 16 18 20 22 24
1
2
3
4
5
time [h]
oven.count
freezer.count
0 2 4 6 8 10 12 14 16 18 20 22 24
0
10
20
30
40
50
60
time [h]
Energy [kWh]
Electrical Grid Energy
Cooling Grid Energy
Heat Grid Energy
This demonstrates that interdisciplinary energetic modeling
of production facilities is feasible with the proposed hybrid
simulation approach using the Hybrid PDEVS formalism, and
can exploit dynamic effects across the whole model. The
simulation results facilitate quantified predictions about ener-
getic behavior in different production scenarios and operating
strategies and display effects that demonstrate the potential for
systematic optimization.
VI. DISCUSSION
The main achievement of the proposed approach is the
improved modularity of cube models, which is attributed to the
ability to create hybrid model components. This significantly
increases reusability of the produced artifacts compared to
other approaches, thereby simplifying and accelerating future
model development, especially for interdisciplinary models
that span multiple domains of engineering.
A disadvantage we encountered is the “roughness” and
generecity of the underlying formalism, i.e. the modelers must
address basic simulation aspects of their application domain
when building a model (see Section IV-C). This is time-
consuming during initial model implementation and facilitates
subtle inconsistencies between model blocks, which might
cause incompatibilities. Further, no high-level software tools
are available yet that better support applied modeling using
Hybrid PDEVS as such DEVS-related formalisms are popular
mostly in academia. This correlates with the difficulty to trans-
fer models, or their descriptions, between tools and/or engines,
due to the lack of established model description languages,
although some remarkable efforts are observable [18].
Directly following the prototype implementation, investiga-
tions regarding integration into existing automation systems
could be initiated, aiming at offering customers automated
simulation-based prediction and optimization functionality
for operation and control of production facilities. Software
providers implemented a comparable simulation engine and
cube models and verified their results using the presented
MATLAB implementation. First performance evaluations ex-
ceeded our expectations regarding simulation speed. This
indicates a sufficiently large headroom for heuristic opti-
mization techniques, e.g. pattern search or population based
algorithms. Such an algorithm could for example search for
optimal production schedules by minimizing overall energy
demand, taking into account energy infrastructure and building
conditioning. Various additional exploitation possibilities of
the BaMa methodology also arose from this joint venture.
VII. CONCLUSION AND OUT LO OK
We successfully demonstrated how the BaMa cubes im-
plemented on a Hybrid PDEVS simulator offer a feasible
and well-performing alternative to the usual approaches when
tackling large and complex, interdisciplinary systems. Addi-
tionally, we shared our findings and the techniques we applied,
when working with a generic and low-level formalism.
In the course of BaMa, a working simulation component
enables systematic simulation-based prediction and optimiza-
tion of energy and resource demands while considering the
other economic success factors costs, time and quality.
The next step is to evaluate optimization techniques and
which algorithms perform well for substantially different mod-
els and how the surrounding conditions affect the optimization,
e.g. how and when boundary conditions may be handled.
Further, a framework that eases interconnection between sim-
ulation model and optimization framework shall be defined.
Based on the findings explained in Section IV-C, it would
be desirable to define a abstraction layer on top of Hybrid
PDEVS which incorporates reoccurring modeling techniques
in a standardized manner, e.g. entity exchange and buffering
of inputs, so that the modeling of the respective blocks could
be drastically simplified.
ACK NOW LE DG EM EN T
We want to thank all partners of the BaMa project for their
contributions. The research presented is funded by the Austrian
Climate and Energy Funds within the program e!MISSION.at
– Energy Mission Austria, project number 840746.
REFERENCES
[1] Z. Irani, V. Hlupic, L. P. Baldwin, and P. E. Love,
“Re-engineering manufacturing processes through sim-
ulation modelling,” Logistics Information Management,
vol. 13, no. 1, pp. 7–13, 2000.
[2] C. Herrmann, S. Thiede, S. Kara, and J. Hesselbach,
“Energy oriented simulation of manufacturing systems–
concept and application,” CIRP Annals-Manufacturing
Technology, vol. 60, no. 1, pp. 45–48, 2011.
[3] I. Leobner, W. Mayrhofer, B. Heinzl, I. Kovacic, P.
Smolek, and K. Ponweiser, “Balanced Manufacturing
– a methodology for energy efficient production plant
operation,” in Proceedings of 10th Conference on Sus-
tainable Development of Energy, Water and Environ-
ment Systems, Dubrovnik, Sep. 2015.
[4] F. Preyser, I. Hafner, and M. R¨
oßler, “Implementation
of hybrid systems described by dev&dess in the qss
based simulator powerdevs,SNE, p. 109,
[5] C. Deatcu and T. Pawletta, “A qualitative comparison of
two hybrid devs approaches,SNE - Simulation Notes
Europe, vol. 22, no. 1, pp. 15–24, 2012.
[6] A. Chow and B. Zeigler, “Parallel devs: a parallel,
hierarchical, modular modeling formalism,” in Proc.
Winter Simul. Conf., IEEE, Dec. 1994, pp. 716–722.
[7] B. Heinzl, W. Kastner, I. Leobner, F. Dur, F. Bleicher,
and I. Kovacic, “Using coupled simulation for planning
of energy efficient production facilities,” in 2014 Work.
Model. Simul. Cyber-Phys. Energy Syst., Apr. 2014.
[8] M. Wetter, “Co-simulation of building energy and con-
trol systems with the building controls virtual test bed,
Journal of Building Performance Simulation, vol. 4, no.
3, pp. 185–203, Sep. 2011, IS SN : 1940-1493.
[9] J. S. Dahmann, R. M. Fujimoto, and R. M. Weatherly,
“The department of defense high level architecture,” in
Proceedings of the 29th conference on Winter simula-
tion, IEEE Computer Society, 1997, pp. 142–149.
[10] B. M. Kelley, P. Top, S. G. Smith, C. S. Woodward,
and L. Min, “A federated simulation toolkit for electric
power grid and communication network co-simulation,
in Modeling and Simulation of Cyber-Physical Energy
Systems (MSCPES), 2015 Workshop on, IEEE, 2015.
[11] E. Widl, B. Delinchant, S. Kubler, D. Li, W. Muller,
V. Norrefeldt, T. Nouidui, S. Stratbucker, M. Wetter,
F. Wurtz, and W. Zuo, “Novel simulation concepts for
buildings and community energy systems based on the
functional mock-up interface specification,” in Model-
ing and Simulation of Cyber-Physical Energy Systems
(MSCPES), 2014 Workshop on, Apr. 2014, pp. 1–6.
[12] R. Goldstein, S. Breslav, and A. Khan, “Informal devs
conventions motivated by practical considerations,” in
Symp. Theory Model. Simul. - DEVS Integr. M&S Symp.
(DEVS 13 ), San Diego, CA: ACM, 2013.
[13] B. P. Zeigler, H. Praehofer, and T. G. Kim, Theory of
modeling and simulation: integrating discrete event and
continuous complex dynamic systems. Academic press,
2000.
[14] H. Pr¨
ahofer, “System theoretic foundations for com-
bined discrete-continuous system simulation,” disserta-
tion, VWG ¨
O, Vienna, 1992.
[15] E. Kofman and S. Junco, “Quantized-state systems:
A devs approach for continuous system simulation,
Transactions of the Society for Modeling and Simulation
International, vol. 18, no. 3, pp. 123–132, 2001.
[16] F. E. Cellier and E. Kofman, Continuous system simu-
lation. Springer Science & Business Media, 2006.
[17] T. Pawletta, C. Deatcu, S. Pawletta, O. Hagendorf, and
G. Colquhoun, “Devs-based modeling and simulation
in scientific and technical computing environments,
Simul. Ser., vol. 38, no. 1, p. 151, 2006.
[18] M. Nikolaidou, V. Dalakas, L. Mitsi, G.-D. Kapos,
and D. Anagnostopoulos, “A sysml profile for classical
devs simulators,” in Software Engineering Advances,
2008. ICSEA’08. The Third International Conference
on, IEEE, 2008, pp. 445–450.
... Parts of this thesis have also been published in other papers over the course of multiple research projects. The most relevant ones are [136,135,229,130,252,133,132,129,28,29,134]. 7 ...
... After evaluating both methods with regard to their capabilities to model hybrid systems in an industrial context [222], it turned out that DEV&DESS was unfit for our applications, due to its shortcomings related to Classic DEVS, and we instead had to opt for a solution based on Parallel DEVS (PDEVS), i.e. hyPDEVS. The drawback, however, is that hyPDEVS is less known in academia and is lacking off-the-shelf tool support [229]. 50 ...
... The case study is a simplified model of a real production plant of an industrial bakery that produces baked goods [229]. The conceptual model of this case study is depicted in Figure 5. 10. ...
... The building-related cubes are further explained in Building model within the BaMa digitaltwin ecosystem; however, a detailed description of all other cube models would be beyond the scope of this study. Further information is available in Raich et al. (2016), Smolek et al. (2017), and Smolek et al. (2018). ...
... A set of use cases has been used to define building-related information exchange requirements, along with information about TBS, production processes, and logistics, needed for the hybrid simulation models. First simple prototypes were developed to test modular cubes (Raich et al., 2016;Smolek et al., 2018), which then evolved into models of real manufacturing facilities from project partners Smolek et al., 2017). For the domain of BEM inside the hybrid simulation, necessary input for the two related cube types, as described in Building model within the BaMa digital twin ecosystem and Figure 4, was identified based on traditional requirements of BEM tools assorted appropriately to the cube approach. ...
Article
Energy and resource efficiency as well as reduction of emissions are nowadays significant objectives for production companies. Industry 4.0, through extensive digitalization along the value chain, enables the achievement of these objectives not only in the construction of new facilities but also in existing facilities as well. This requires an interdisciplinary approach, extending over production and logistic processes as well as the building, technical building services, and energy supply systems, consolidated through integrated modeling and simulation-based optimization. The research question this study addresses is how to digitally couple these subsystems and optimize the overall system’s performance in terms of energy and resource efficiency, by distancing from silo-field thinking while using an integrated analysis approach. The article briefly presents a holistic modeling and simulation framework, utilizing modular digital twins (DTs) of all elements that may constitute a given industrial unit. The integration of multiple DTs of these subsystems in a hybrid (continuous and discrete) simulation forms a holistic DT ecosystem of an existing facility. The particular focus of the study is the building representation in this DT ecosystem for energy-efficient production. Based on a methodology including hybrid simulation, building information modeling (BIM), and visual programming, a semi-automated data acquisition workflow was proposed. The hybrid simulation is based on Discrete Event System Specification (DEVS) formalism, where the building is incorporated as a building energy model (BEM). Within the abstracted representation of the overall system, the article explores the possibilities of parametrizing the DT of the building, interconnected with the rest of the factory elements, by acquiring information directly from existing BIM models. Through a comparative case study, the proposed workflow is compared to a manual one in terms of integrity and benefits. The study’s contribution lies in: 1) the detection of the required building level of abstraction for a holistic DT ecosystem, 2) the definition of the interconnections between the building-related counterparts and the rest of the virtual environment as well as the data required for their parameterization, and 3) proposing a semi-automated workflow via virtual programming, for BIM-based creation of the building model within a holistic DT ecosystem.
... For these reasons, we employ a formal model description, called hyPDEVS [39] for hybrid system modelling [22]. Compared to using typical co-simulation methods, hyPDEVS offers tighter hybrid integration and improved modularity [36] by following a strict component-based paradigm that defines atomic and coupled components, which can be combined to create new application models. ...
... The components are designed for reusability and take into account entity exchange as well as energy balance equations. More details on the hybrid simulation are described in [39,42]. ...
Article
Full-text available
This paper presents the development and evaluation of a digital method for multi-criteria optimized production planning and control of production equipment in a case-study of an Austrian metal casting manufacturer. Increased energy efficiency is a major requirement for production enterprises, especially for energy intensive production sectors such as casting. Despite the significant energy-efficiency potential through optimized planning and the acknowledged application potential for sophisticated simulation-based methods, digital tools for practical planning applications are still lacking. The authors develop a planning method featuring a hybrid (discrete-continuous) simulation-based multi-criteria optimization (a multi-stage hybrid heuristic and metaheuristic method) for a metal casting manufacturer and apply it to a heat treatment process, that requires order batching and sequencing/scheduling on parallel machines, considering complex restrictions. The results show a ~10% global goal optimization potential, including traditional business goals and energy efficiency, with a ~6% energy optimization. A basic feasibility demonstration of applying the method to synchronize energy demand with fluctuating supply by considering flexible energy prices is conducted. The method is designed to be included in the planning loop of metal casting companies: receiving orders, machine availability, temperature data and (optional) current energy market price-data as input and returning an optimized plan to the production-IT systems for implementation.
... However, as discussed in [5], the use of transitory states leads to a delay of events regarding processing order, which in turn impedes reusability of components. Due to the reasons mentioned above and the experiences we made with applying both, DEVS [6] and PDEVS [7], we decided to revise PDEVS resulting in RPDEVS published in [8]. Basically, the changes include the support of 'true' mealy behavior and the merging of the three state transition functions δ int , δ ext , and δ con f into one generic state transition function δ . ...
... However, as discussed in [5], the use of transitory states leads to a delay of events regarding processing order, which in turn impedes reusability of components. Due to the reasons mentioned above and the experiences we made with applying both, DEVS [6] and PDEVS [7], we decided to revise PDEVS resulting in Revised Parallel DEVS (RPDEVS) published in [8]. Basically, the changes include the support of 'true' mealy behavior and the merging of the three state transition functions δ int , δ ext , and δ con f into one generic state transition function δ . ...
Conference Paper
Full-text available
The Revised Parallel DEVS (RPDEVS) model-ing formalism enhances the Parallel Discrete Event System Specication (PDEVS) by the ability to model 'real' Mealy behavior of components. The term 'real' Mealy behavior can be summarized as immediate output response to an input event without a state transition in between. Although this enhancement simplies model creation, especially of reusable components, it requires a more complex simulation algorithm. In this paper, we present an RPDEVS abstract simulator that describes the simulation execution of RPDEVS models.
... For the continuous aspects, energy balance equations model the transient dynamics of the internal temperature. A detailed explanation of the oven Cube as well as a translation of the oven model into hyPDEVS are given in Ref. [18]. ...
Article
Full-text available
An energy-efficient production is imperative and can reduce costs. Despite the acknowledged potential to increase energy efficiency in production systems through production planning and control (PPC), adequate planning methods are lacking. This article presents an interdisciplinary approach for a simulation-based multi-criteria optimization, integrating energy efficiency into PPC objectives. The method considers production equipment together with HVAC and technical building services. It features a novel integrated hybrid discrete/continuous simulation method enabling to accurately capture dynamic interactions between material and energy flows. The approach is evaluated in a case study on the food industry, indicating potential energy efficiency gains of up to 30%.
Book
Full-text available
Effective and efficient planning of energy efficiency measures is of great importance to manufacturing companies. Material flow simulation that has been extended to also include energy is increasingly used in this context as a tool for the analysis of complex interactions between material flows and energy flows in factories. This thesis deals with the conception and the systemization of a methodology for designing such simulation studies specifically in re-planning projects that aim for energy efficiency improvements. Taking basic approaches from decision theory into particular consideration, it is intended to provide guidance in deciding on the project-specific manifestation for relevant characteristics of a simulation study in this problem area (e.g. the manner to model energy consumption), regardless of the utilised simulation solution. The developed solution comprises a 13 steps spanning process model as well as nine solution modules to support decisions concerning the choice of manifestation for selected characteristics. In this way, the entire process from the development of an energy efficiency measure through the actual application of simulation (following VDI guideline 3633 Part 1:2014) to the eventual investment decision is assisted. The results of this thesis were exemplarily tested in three case studies. All initially defined requirements could thereby be positively verified.
Article
In this work, we present a Revised Parallel DEVS (RPDEVS) formalism. The Classic Discrete Event System Specification (DEVS) and Parallel DEVS (PDEVS) formalisms do not support modelling of ‘true’ mealy behaviour, i.e. reacting to an input message immediately with an output message. Instead, such behaviour has to be modelled via transitory states and multiple state updates. This not only increases model complexity, it also impedes reusability of model components in different contexts. RPDEVS enhances PDEVS with the capability to model mealy behaviour directly. Hence, the output function λ can access the input bag. This introduces some challenges regarding the simulation algorithm which we will take a look at. Further, the terms algebraic loop and illegitimate model will be discussed in the context of RPDEVS. It is shown that RPDEVS models which are free of algebraic loops are also legitimate. Finally, it will be demonstrated that like Classic DEVS and PDEVS, also RPDEVS provides closure under coupling.
Conference Paper
Full-text available
This paper describes our current research in the area of Discrete Event System Simulation (DEVS) and its implementation for programmable Scientific and technical Computing Environments (SCEs) with a focus on variablestructure and hybrid systems. Engineers, unlike scientists, are usually familiar with the use of SCEs such as Matlab rather than high level programming language simulation libraries. DEVS-based modeling and simulation until now has not been available for SCEs. This research has led to the development of a fully compatible toolbox for the Matlab environment with the potential to interact with other toolboxes. The paper reviews the advantages of DEVS/SCEs integration and concludes by describing the potential benefits in applications with other toolboxes.
Conference Paper
Full-text available
In this article, a method for implementing hybrid models, formulated as DEV&DESS, in the QSS based simulator PowerDEVS is presented. PowerDEVS is actually a pure DEVS simulator. However, it is specialised for simulating continuous Models (DESS) using QSS (Quantized State System) to translate them into discrete event models (DEVS). Therefore, it is perfectly suited for the simulation of hybrid models (DEV&DESS). When designing and simulating coupled DEVS models though, very soon a lot of difficulties occur caused by concurrent events and feedback loops. Hence, first some concepts are introduced of how to implement an atomic DEVS in a way that avoids those difficulties. Afterwards, an approach of how to implement an atomic DEV&DESS is introduced which makes use of those concepts.
Conference Paper
Full-text available
Research and development in the fields of building technologies and community energy systems have caused in recent years a transition away from stand-alone components towards dynamically interacting systems. However, the precise modeling and simulation of such complex cyber-physical systems proves challenging for the established simulation tools. This paper illustrates the applicability of co-simulation and model exchange concepts based on the Functional Mock-up Interface (FMI) specification for the simulation of buildings and community energy systems by comparing several state-of-the-art approaches. The presented applications thereby demonstrate the suitability and relevance of such modular and flexible simulation concepts for these fields. At the same time the importance of a common simulation interface is emphasized, which allows the reuse of individual components for a diverse range of applications.
Article
Full-text available
A hybrid system is defined as a system with mixed discrete event and continuous parts. Two approaches for modeling and simulation of hybrid systems in the context of the Discrete EVent System Specification (DEVS) formalism are compared. Since the system theoretic DEVS approach has been introduced during the 1970ies, it has been completed by different extensions. The hybrid extensions checked against each other here, are the Quantized State Systems (QSS) method and a wrapper concept based on traditional ODE solvers. For a comparison from engineering practice point of view, these two hybrid DEVS approaches are analyzed within the scientific and technical computing environment Matlab.
Conference Paper
This paper introduces a federated simulation toolkit (FSKIT) that couples continuous time and discrete event simula- tions (DES) to perform the co-simulation of electric power grids and communication networks. A High Performance Computing (HPC) oriented power system dynamic simulator, GridDyn, was used for the electric power grid simulation. GridDyn is coupled to the open-source network simulator, ns-3, through FSKIT. FSKIT provides time control for advancing the state of federated simulators, and facilitates communication among objects in the federate. A wide-area communication-based electric transmission protection scheme is simulated with FSKIT, using the IEEE 39- bus test system. A communication network for the 39-bus system is built in ns-3, and basic protection relay logic is added to the power system model in order to perform the co-simulation.
Article
Continuous System Simulation describes systematically and methodically how mathematical models of dynamic systems, usually described by sets of either ordinary or partial differential equations possibly coupled with algebraic equations, can be simulated on a digital computer. Modern modeling and simulation environments relieve the occasional user from having to understand how simulation really works. Once a mathematical model of a process has been formulated, the modeling and simulation environment compiles and simulates the model, and curves of result trajectories appear magically on the user's screen. Yet, magic has a tendency to fail, and it is then that the user must understand what went wrong, and why the model could not be simulated as expected. Continuous System Simulation is written by engineers for engineers, introducing the partly symbolical and partly numerical algorithms that drive the process of simulation in terms that are familiar to simulation practitioners with an engineering background, and yet, the text is rigorous in its approach and comprehensive in its coverage, providing the reader with a thorough and detailed understanding of the mechanisms that govern the simulation of dynamical systems. Continuous System Simulation is a highly software-oriented text, based on MATLAB. Homework problems, suggestions for term project, and open research questions conclude every chapter to deepen the understanding of the student and increase his or her motivation. Continuous System Simulation is the first text of its kind that has been written for an engineering audience primarily. Yet due to the depth and breadth of its coverage, the book will also be highly useful for readers with a mathematics background. The book has been designed to accompany senior and graduate students enrolled in a simulation class, but it may also serve as a reference and self-study guide for modeling and simulation practitioners. © 2006 Springer Science+Business Media, Inc. All rights reserved.
Conference Paper
To meet the increasing requirements of sustainable production and to assure economic competitiveness, novel strategic approaches are necessary in designing production systems and to increase energy efficiency in the manufacturing industry. Individual areas within production facilities (e.g. production system, energy system, building hull) can be analyzed individually by using simulation-based methods. In order to access additional optimization potential, it is necessary to expand the boundaries of such simulations and to consider dynamic interactions between individual optimization fields. This work presents an approach for an interdisciplinary co-simulation, in which, for an overall integrated simulation, several simulation environments are coupled that periodically exchange data quasi-parallel at runtime. This allows not only to combine different model descriptions, but also multiple calculation algorithms, each specifically tailored to the individual needs of the respective field of engineering. A case study of a metal-cutting production facility presents an application example. The goal is to provide a decision-support tool for the early planning stages of production facilities that allows making qualified predictions about the effect and financial impact of different energy saving measures.
Article
A new class of dynamical systems, Quantized State Systems or QSS, is introduced in this paper QSS are continuous time systems where the input trajectories are piecewise constant functions and the state variable trajectories-being themselves piecewise linear functions-are converted into piecewise constant functions via a quantization function equipped with hysteresis. It is shown that QSS can be exactly represented and simulated by a discrete event model, within the framework of the DEVS-approach. Further, it is shown that QSS can be used to approximate continuous systems, thus allowing their discrete-event simulation in opposition to the classical discrete-time simulation. It is also shown that in an approximating QSS, some stability properties of the original system are conserved and the solutions of the QSS go to the solutions of the original system when the quantization goes to zero.
Article
Energy efficiency has become a key concern in industry due to the increased energy costs and the associated environmental impacts. Besides considering single processes/machines, the perspective on process chains and factories as a whole bears further potentials for improvement. In this context, dynamic interactions of different processes as well as auxiliary equipments (e.g. compressed air generation) need to be taken into account when planning and controlling manufacturing systems. This paper presents an innovative energy oriented simulation model for planning of manufacturing systems. A number of industry cases are used to demonstrate the potentials and applicability of the proposed methodology.