Available online at www.sciencedirect.com
Procedia Manufacturing 00 (2017) 000–000
2351-9789 © 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the 15th Global Conference on Sustainable Manufacturing.
15th Global Conference on Sustainable Manufacturing
Hybrid simulation-based optimization of discrete parts
manufacturing to increase energy efficiency and productivity
Thomas Sobottkaab*, Felix Kamhuberab, Matthias Rösslerc, Wilfried Sihnab
aFraunhofer Austria Research GmbH, Theresianumgasse 7, Vienna 1040; Austria
bVienna University of Technology, Theresianumgasse 27, Vienna 1040; Austria
cdwh GmbH, Simulation Services, Neustiftgasse 57-59, Vienna 1070; Austria
* Corresponding author. Tel.: +43-676-888-61626; E-mail address: email@example.com
This presented research comprises the development of an optimization module for use in a novel production optimization tool –
similar in function but not mode of operation to an Advanced Planning System –, with energy efficiency incorporated into its
goal system. The optimization features a hybrid-simulation of production systems as an evaluation function. A hybrid simulation
has been developed and presented in preceding publications, in order to enable a sufficient consideration of interactions between
material flow and the thermal-physical behavior of the production system. The size of the search space for the complex
optimization problem necessitates a customized two-phase-optimization method, which is based on a Genetic Algorithm, with the
consideration of linear constraints and extended customizations. The results, obtained in a case study featuring a food production
facility, show energy savings of around 20 percent together with significant productivity gains.
© 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the 15th Global Conference on Sustainable Manufacturing.
Keywords: sustainable manufacturing; energy efficiency; production planning; optimization; metaheuristics; hybrid simulation; case study
Human contribution to climate change necessitates urgent action: In the United Nations Framework
Convention on Climate Change (UNFCCC)  countries worldwide have committed to limiting the manmade
temperature rise to 2°C, mainly by reducing energy related CO2 emissions by 1.9% between 2013 and 2040. The
increased awareness and societal pressure for increased sustainability, together with long-term trends of rising
energy costs and the fact that the industrial sector is responsible for 31% of the annual energy demand and 36% of
the CO2 emissions globally , are turning energy efficiency into a substantial goal for manufacturing enterprises
2 Author name / Procedia Manufacturing 00 (2017) 000–000
. The basic definition of energy efficiency in the context of production is the ratio between the value-added
output of a production system and the necessary energy input .
In addition to the need to be more energy efficient, there are also chances for companies in the domain of energy
efficiency and sustainable manufacturing: The ongoing energy transition towards renewable energy sources leads to
a more volatile energy supply. The larger the share of renewable energy in the supply mix, the better the demand has
to be aligned with the energy production – bidirectional approaches such as demand response are one way of
achieving this . This also means that if companies can predict, plan and control their energy consumption profile
according to changing supply situations, they can benefit from very low energy prices on short-term energy markets.
Both goals, increasing energy efficiency and the ability to plan and control the energy consumption profile,
necessitate advanced planning tools for companies, especially for manufacturing companies with complex
productions systems. Studies show that energy efficiency considerations should be an integral part of Enterprise-
Resource-Planning (ERP) and Manufacturing Execution Systems (MES), with simulation based approaches
suggested to be the most promising method. However, currently there is a lack of practically applicable planning
In order to address this deficiency, this research is meant to develop a novel planning tool that increases both the
energy efficiency and general performance of production systems, using a hybrid simulation-based optimization
approach. The general planning concept has been published by the research team , so has the hybrid simulation
concept  and the development of the optimization module . The particular paper at hand focuses on the
adaption of the planning method to a specific real life industrial application and on evaluating the optimization
potential in an industrial use case.
2. Requirements for the Planning Tool and chosen Approach
The requirements for a planning tool have been deducted from an analysis of relevant literature, supplemented
by expert interviews with managers from application partners within the research project. According to the findings
of a large EU research project, based on interviews with 106 international experts, the tools will have to consider
both conventional economic planning goals and energy and resource efficiency simultaneously. The planning should
be integrated into the existing ICT and the use of detailed simulation models is recommended . Li  stresses
the necessity for a generalized structure to make the tools available for different application environments. He 
also emphasizes the need to consider all relevant energy flows and their interdependencies. Concerning the
underlying methods, Thiede – among others – declares simulation to be the method best suited to provide the
necessary planning support . An automatic decision support function, i.e. in the form of an optimization module,
is another major request from prospective industry users. A classic decision support function for production
planning and control in company Information and communications technology (ICT) is found in Advanced Planning
Following a comprehensive evaluation of possible approaches for this research, a simulation based optimization
emerged as the most suitable option: It enables a planning functionality similar to that of an APS, due to the
automatic optimization of production schedules and control of machines in the production process and the periphery.
It also enables the simultaneous consideration of the complex system of material flows and the thermal-physical
behavior of the production and energy system and its components through a simulation of said production system.
3. Related Work
3.1. Simulation & simulation based Planning approaches
The basic concept of a dynamic simulation is to create a digital model of a real-life system, featuring all relevant
characteristics, and to then use that model to conduct experiments in order to gain insights into the system behavior
or to optimize and develop plans for the system . Typically, material flows and the processing of orders are
simulated utilizing a discrete event simulation (DES), while the thermal-physical behavior of machines and
equipment is usually simulated in continuous simulation environments that solve ordinary differential equations
(ODEs) or differential-algebraic equations (DAEs). One of the most advanced simulation based concepts for
Author name / Procedia Manufacturing 00 (2017) 000–000 3
planning tools in the field of energy aware planning is Thiede’s approach  based on AnyLogic® in a multilevel-
simulation [15, 16]. This concept combines multiple simulation environments in a co-simulation, in which the
subsystems are modelled either in a DES or DESS environment. The sub-simulations are coupled at certain points
during the simulation run and provide a certain level of integrated modelling. In order to ensure the cooperation of
different model parts in a co-simulation, the respective solvers of the subsystems have to be synchronized. This
generates various iterations and overhead, especially for solvers with a variable step size, which are commonly used
in continuous simulation environments. As the focus of this research is more on the continuous part than Thiede’s
work, a co-simulation approach is not feasible and an integrated hybrid simulation was chosen and developed,
enabling the modeling of continuous and discrete behavior simultaneously.
3.2. Optimization & simulation based optimization
Optimization methods comprise optimization algorithms that are able to find optimal solutions for problems with
a limited complexity and optimization heuristics that are able to find approximate solutions, if exact solutions cannot
be found; these are called NP-hard problems. Within heuristics, there are special heuristics dedicated to a certain
problem category in operations management and metaheuristics that serve as generic algorithms for a broader range
of applications when dedicated heuristics are not available . Most complex optimization tasks, especially if the
optimization utilizes a simulation as an evaluation function, as in this research, feature multiple local optima. This
requires the algorithms to not „get stuck” in local optima, in order to arrive at better solutions eventually . The
metaheuristics can be discerned in algorithms based on iterative local search (LS) and generative population based
methods (PS) . Most of these heuristics mimic natural processes, i.e. imitating animal behavior, evolution in
biology or the cooling process of materials.
The optimization problem in this research is mainly the scheduling and sequencing of orders, thus representing a
permutation flow shop sequencing problem (PFSSP), extended by the optimal control of production equipment and
equipment in the periphery. For optimization in production scheduling applications without the energy aspect,
Pochet gives an overview of approaches based on mixed integer programming . Due to the increased model
complexity in the case of energy aware planning, there are examples of approaches based on the Genetic Algorithm
(GA) . Rager  utilizes the GA for a simulation based approach in a case similar to the projected application
of this paper. None of the existing optimization methods features and supports the more complex hybrid simulation
developed and used in this research, thus a customized optimization method has to be developed.
4. Hybrid Simulation
In order to be used within the proposed planning tool the simulation has to fulfil three major requirements:
Support for hybrid models: The combination of logistics simulation, which is typically simulated using a discrete
event simulation (DES), and the simulation of the energy flows, that is continuous in nature and are therefore
modelled as ordinary differential equations (ODEs) or differential algebraic equations (DAEs). This combination
is not supported by current planning tools and thus is the main innovation of the proposed software tool.
Modular structure of models: For the use in a wide variety of settings, the models for the simulation should
feature a modular structure. This enables the development of a library of basic components of production
facilities (such as belts, ovens, manufacturing equipment, HVAC-systems, thermal zones) that can be used to
build the model of the facility and that can be used for the simulation.
Description in one formalism: The integration of the simulation in an existing MES/APS can be realized by either
connecting to an existing, external system or implementing the simulation environment in the program itself. For
this project the second method was chosen. As mentioned before, the usage of a co-simulation approach is not
feasible for the focus of this research. Thus, the models have to be described in one formalism in order to keep
the effort for development and implementation of the simulation engine as low as possible.
4 Author name / Procedia Manufacturing 00 (2017) 000–000
In order to fulfill these requirements the DEV&DESS-formalism in combination with PDEVS was chosen to
describe the hybrid models – in the context of this research the approach was called “Hybrid P-DEVS”. For testing
purposes and first implementations the MATLAB-DEVS-Toolbox developed by HS Wismar  was used. For the
purpose of the optimization, the performance of the simulation had to be tweaked, so a software implementation
partner implemented the engine in C++ .
5. Optimization Module
The optimization module that utilizes the hybrid simulation as an evaluation function was developed in two
phases: First, different metaheuristics that are suitable for complex multimodal (featuring local optima) fitness-
landscapes were evaluated using identical scenarios. Second, the best performing heuristics from phase one were
then enhanced and customized to provide an optimal fit for the given optimization task and model behavior.
At the end of the development process, a Genetic Algorithm (GA), with a set of tuning and customization
measures for optimal optimization performance, emerged as the best performing solution. In a simplified test case,
the optimization was able to improve the objective function of the optimization by up to 30%. More details by the
same authors concerning the optimization module have been published .
6. Potential Analysis in a Case Study
6.1. Case-Study Setup
For this Case Study, the simulation of an actual production line for rolls in an industrial bakery plant was chosen
as the production system. It consists of nine major production machines, nine conveyor belts with junctions and two
storage components within the production logistics system. The products, baked and deep-frozen rolls, use different
material flow variants – mainly with and without passing through an industrial oven – and require different process
parameters, e.g. temperatures and processing times on machines. These product characteristics are stored on and
used as input via process sheets. Two of the production machines – an industrial oven and a freezer – feature a
distinct thermal-physical behaviour. The basic model structure is depicted in Fig. 1.
Fig. 1. Simulation model: structure plus material and energy flows
In the Technical Building Services (TBS) and energy system, there is a heater providing heat via a heat network
to the industrial oven and the different halls/subsections of the factory building. Next to the heat network, there are
Author name / Procedia Manufacturing 00 (2017) 000–000 5
three cold networks and one electric network. The cold networks contain five chillers. The external environment is
considered by importing weather data – i.e. temperature, wind and cloud coverage. Scenarios for three seasons in
2016 have been tested (seasons following the weeks 03/34/44). In the simulation model, the building itself is divided
into four thermal zones, representing the main production hall, a room with technical building services equipment,
an office building and a freezer warehouse – each of those zones is supplied with heat/cold and exchange heat with
each other during the simulation. The hybrid simulation is implemented in a C++ software and the optimization is
implemented in Matlab® using the Global Optimization Toolbox. The simulation model was parametrized with
actual data from production, process and comprehensive energy measurements conducted in the production plant.
The production scenarios presented in this paper feature 1-day-/7-day-scenarios (simulated time) and feature all
12 product types that are produced on the production line. The imported production/demand schedules are actual
production schedules from the fall of 2016. The base planning is the actual production schedule created by manual
planners that was executed in the past against which the optimization results are now being evaluated.
Concerning the multi-criteria objective function for the optimization, the basic function developed in  was
adapted – i.e. parametrized and weights assigned to the part objectives – in the course of goal-system workshops
with the planners and management of the bakery. The following function emerged:
- part-goal weights
… index for production lots
… evaluation function for late deliveries and storage costs
… lot completion time
… accumulated overall energy costs
… overall energy cost per kWh
… number of unfinished goods (demanded goods not produced)
… number of unfinished goods (goods not produced)
The objective function is normalized to calculate a cost value denominated in Euro. The part-function
calculates storage costs for goods finalized before the delivery date and penalty costs for delayed deliveries as a
function of the order completion time. The second part considers the energy consumption – the overall energy costs
are calculated as a combination of the actual energy cost per energy source and a cost value for the associated CO2
emissions. For the actual electric energy costs, a number of scenarios have been tested: Season 0 features the current
fixed price for the bakery, whereas scenarios with seasons 1/2/3 use the actual spot-market prices for electricity in
the winter/summer/autumn of 2016 (seasons following the weeks 03/34/44) . Thus, in the latter three scenarios
the opportunities of a complex energy-portfolio-management utilizing the short-term markets are evaluated. The
energy cost for cooling is calculated using the consumption data of the electrically powered coolers. For the heat
production the actual costs for the company was used. Concerning the cost value for the associated CO2 emissions,
the energy usage per energy source was combined with corresponding conversion rates for Austria , evaluated
with Spot-market prices of the European Energy Exchange AG  and an additional value, which the management
of the bakery assigned to reducing CO2 emissions. The third part of the objective function evaluates penalty costs
for unsatisfied demand during the simulation and the fourth penalizes unfinished goods at the end of the simulation.
6.2. Adaptations to the Optimization Module
The optimization module developed in  has been adapted for the complex production system of the case study.
The major components of the optimization, a customized/tuned Genetic Algorithm (GA), remain unchanged. The
customizations comprise: a guided search by adapted operators in the GA, a memory function from the Tabu
Search algorithm, a mixed integer optimization (default value 1), hybridization by combining the GA with the PS
and determining the optimal population size (the population size was set to 12). The most prominent adaptation of
this basic optimization is the separation of the optimization procedure into two phases – this is a measure to improve
6 Author name / Procedia Manufacturing 00 (2017) 000–000
the runtime of the optimization. Due to the large search space created by real life production systems, it proved
beneficial to only enable the sequencing and scheduling of orders in the first phase. This is followed by a second
phase with a fixed order sequence, during which the operation time windows for machines and equipment in the
periphery are being shifted and contracted, mainly in order to minimize the energy consumption. It is important to
note however, that although the actuation variables of the optimizations change, the simulation is always
considering the entire production and energy system. Thus, in both phases the entire objective function can be
optimized, including the energy consumption. The new approach requires adaptations in the constraints in both
phases. In the first phase, the only necessary constraint is to have positive order release times. In the second phase,
overlapping processing on any given machine has to be prevented as well as ensuring the operating times on every
machine are at least as long as the minimum processing time for each production lot on the same machine.
A second major adaption is the introduction of a production plan generator (PPG), the goal of which is to
minimize the number of practically infeasible solutions created by the optimization (GA). This in turn significantly
decreases the number of necessary, computation-intensive, simulation evaluations and prevents the GA from
“getting lost” in search space areas without practically – according to the objective function – permissible solutions.
The PPG accesses the process sheets and, based on received order release times, calculates production plans for each
machine and piece of equipment. The PPG significantly increases the optimization speed and quality of the results.
In phase 1 the PPG is called before every simulation run for every intermediate solution. Between phases 1 and 2 it
is called again to ensure feasible oven operating times and ensuring the linear constraints are not violated. In phase 2
the PPG also receives optimized oven times for every intermediate solution to calculate the operating times of the
remaining equipment. Given the current simulation performance and the complexity of most real life optimization
tasks with this planning method, the use of a PPG is likely to be necessary for most applications. The downside is
that it has to be adapted to the specific simulation model and thus increases the modelling effort – a fact that the
research team is currently trying to mitigate developing a generalised and modularised approach, including the PPG.
6.3. Case-Study Results
Although scenarios with 1/7/30 days simulated time have been conducted, the most usable anticipated application
case right now is the one day planning horizon, thus the presentation of results will focus on the one day scenarios.
Fig. 2, (a) shows the influence of different seasons and corresponding weather conditions and electrical energy
prices (spot-market) on the optimization potential. Fig. 2, (b) shows the part goal trends during the course of the
optimization by displaying the mean values for each generation of a GA population. The latter diagram shows the
trade-off between the part goals in the objective function and that in phase 2, after a significant improvement in
overall energy costs, the optimization soon reaches a point where no significant improvement is achievable for any
part goal, without creating solutions with unfinished products. Increasing the optimization step-size – a measure to
reduce the possible solution space size – from 1 seconds to 300 seconds improves both optimization speed and
solution quality (see Fig. 3, (a)). It is important to adjust the constraint tolerances to the chosen step size.
Fig. 2. (a) energy goal trend of current best solution; (b) part goal trends of mean values per generation
0500 1000 1500 2000 2500 3000 3500 4000 450 0 5000 5500
goal energy [%]
Scenario 1 (1 day) - goal energy of current best solution
Season 1 Season 2 Season 3
Phase 1 Phase 2
0500 1000 1500 2000 2500 3000 3500 400 0 4500 5000 5500
part goals [%]
goal energy goal dev sho rtage storage
Phase 1 Phase 2
Scenario 1 (1 day) - season 2 - part goal value trend during optimization run
Author name / Procedia Manufacturing 00 (2017) 000–000 7
The overall optimization, in terms of a reduction of the objective function value, in the one-day scenarios was
around 50%. This includes lowering penalties for late deliveries, which are a common occurrence in the manual
planning process. The energy consumption could be lowered by 33%, while the associated overall energy costs –
including CO2 emission costs – dropped by 25%. The scenarios using variable electric energy costs (spot market
prices), showed an increased optimization potential, with the biggest gains in week 34 (summer scenario). This is
due to the larger cooling demand in the summer plus the greater availability of relatively cheaper renewable, solar-
based, energy. The results could be obtained in ~4.400 simulation evaluations and as many intermediate solutions
(with default settings). In the result trends, it is apparent that the number of evaluations to achieve good results could
be reduced to ~2.300 evaluations by shifting the phase transition point to around 2.000 evaluations. On a standard
intel-i7, 4 GHz processor 2.300 simulation evaluations take ~3 hours to execute.
Longer scenarios have been conducted although there are practical limitations: The complexity of seven-day
scenarios, in terms of search space size, is ~300 times larger than that of the one-day scenarios. Given the practical
requirements for an optimization to be executed in not more than 2-3 hours in order to be able to utilize the planning
results, this results in a too small number of possible evaluations to achieve good results, compared to the theoretical
optimization potential. By further reducing the population size, the optimization speed can be significantly
increased, however this also leads to less reliable results – one of the major advantages of population based
metaheuristics is the simultaneous optimization of a large population of intermediate solutions (see Fig. 3, (b)).
Fig. 3. (a) influence of varying optimization step sizes; (b) varying GA population sizes for 7-day scenario
In the next iteration, the optimization module will also be included into the software implementation that already
contains the hybrid simulation. This is expected to significantly improve the overall planning process performance.
With this implementation, the simulation evaluations can be conducted simultaneously, which, together with
common multi-core processors, has the potential to further increase the optimization speed.
7. Conclusions and Outlook
The research presented herein is able to demonstrate that the planning method for the optimized, energy aware
PPC, developed in the course of the research project, is applicable to the planning of real-life production systems. A
number of adaptations within the optimization module were developed in order to cope with the complexity and
specific requirements of the investigated case, while the simulation modelling was possible by relying on the already
developed simulation modules. The results show significant overall optimization potential of up to 50% of the
objective function value, including a reduction of ~30% in the energy consumption. When considering variable
energy prices on spot-markets, the energy-related optimization potential increases. It is important to state that the
optimization potential is very much dependent on specific scenarios – in the majority of the scenarios tested thus far,
the overall potential varied between 15-50%, with the overall energy cost optimization of 8-25%. It also has to be
noted that optimizing the operation time windows of production equipment potentially introduces risk into the
production system. Before implementing schedules created with the planning method, it is therefore advisable to
Scenario 4 (7 days) - global goal of current best solution
popsize 6 popsize 12 popsize 24
goal energy [€]
Scenario 1 (1 day) -int. step-size variation - goal energy, best solution
int 1 int60 int300
Phase 1 Phase 2
8 Author name / Procedia Manufacturing 00 (2017) 000–000
check the generated plans for feasibility; including safety margins to the operation time windows is another possible
measure to mitigate potential risks. The simulation of stochastically distributed risk is not feasible since it would
greatly increase the number of necessary simulation runs, which are already the major limiting factor for the
application of this planning method. The next steps will include a more sophisticated planning for the TBS part of
the production system. While already being simulated with the current tool, the next iteration will include actuating
variables for the TBS, such as a prioritization of multiple devices in one supply network – i.e. coolers – or setting
the target temperature in heat reservoirs and circuits. This will further increase the overall optimization potential and
introduce more complexity for the optimization task. In a final step, the planning method will be applied to
improving entire factories with multiple production groups/lines/plants.
This research is supported by the Climate and Energy Fund as part of the research program Balanced
Manufacturing. The authors would like to thank all project partners.
 United Nations Framework Convention on Climate Change (UNFCCC). Historic Paris Agreement on Climate Change.
 He, Y., Li, Y., Wu, T., Sutherland, J.W., 2015. An energy-responsive optimization method for machine tool selection and operation
sequence in flexible machining job shops 87, p. 245.
 Thiede, S., 2012. Energy efficiency in manufacturing systems. Springer, Berlin.
 Duflou, J.R., Sutherland, J.W., Dornfeld, D., Herrmann, C. et al., 2012. Towards energy and resource efficient manufacturing: A
processes and systems approach 61, p. 587.
 Uslar, M., Specht, M., Dänekas, C., Trefke, J., Rohjans, S., González, J.M., Rosinger, C., Bleiker, R., 2013. Standardization in smart
grids: Introduction to IT-Related Methodologies, Architectures and Standards. Springer, Berlin, Heidelberg.
 Michaloski, J.L., Shao, G., Arinez, J., Lyons, K., Leong, S., Riddick, F. Analysis of sustainable manufacturing using simulation for
integration of production and building service. Society for Computer Simulation International, 2011.
 Popper, N., Hafner, I., Rössler, M., Preyser, F. et al., 2014. A General Concept for Description of Production Plants with a Concept of
Cubes 24, p. 105.
 Raich, P., Heinzl, B., Preyser, F., Kastner, W. Modeling techniques for integrated simulation of industrial systems based on hybrid
PDEVS, in 2016 Workshop on Modeling and Simulation of Cyber-Physical Energy Systems (MSCPES), p. 1.
 Sobottka, T., Kamhuber, F., Sihn, W. Increasing energy efficiency in production environments through an optimized, hybrid simulation-
based planning of production and its periphery, in The 24th CIRP Conference on Life Cycle Engineering.
 Bunse, K., Vodicka, M., Schönsleben, P., Brülhart, M. et al., 2011. Integrating energy efficiency performance in production management
– gap analysis between industrial needs and scientific literature 19, p. 667.
 Li, W., 2015. Efficiency of manufacturing processes: Energy and ecological perspectives. Springer, Cham.
 Herrmann, C., Thiede, S., Kara, S., Hesselbach, J., 2011. Energy oriented simulation of manufacturing systems – Concept and application
60, p. 45.
 VDI. VDI-Richtlinie 3633 Blatt 1, Berlin. Beuth, 2008.
 Thiede, S., Bogdanski, G., Herrmann, C., 2012. A Systematic Method for Increasing the Energy and Resource Efficiency in
Manufacturing Companies: 1st CIRP Global Web Conference: Interdisciplinary Research in Production Engineering (CIRPE2012), p. 28.
 Thiede, S., 2015. Fabrik der Zukunft: intelligent und ressourceneffizient, Wien.
 Schönemann, M., Schmidt, C., Herrmann, C., Thiede, S., 2016. Multi-level Modeling and Simulation of Manufacturing Systems for
Lightweight Automotive Components 41, p. 1049.
 Sivanandam, S.N., Deepa, S.N., 2007. Introduction to genetic algorithms. Springer Science & Business Media.
 Voß, S., Martello, S., Osman, I.H., Roucairol, C., 2012. Meta-heuristics: Advances and trends in local search paradigms for optimization.
Springer Science & Business Media.
 Hertz, A., Widmer, M., 2003. Guidelines for the use of meta-heuristics in combinatorial optimization 151, p. 247.
 Pochet, Y., Wolsey, L.A., 2006. Production Planning by Mixed Integer Programming. Springer Science+Business Media Inc, New York.
 Dai, M., Tang, D., Giret, A., Salido, M.A. et al., 2013. Energy-efficient scheduling for a flexible flow shop using an improved genetic-
simulated annealing algorithm 29, p. 418.
 Rager, M., 2008. Energieorientierte Produktionsplanung: Analyse, Konzeption und Umsetzung. Springer-Verlag.
 Deatcu, C., Pawletta, T. A Qualitative Comparison of Two Hybrid DEVS Approaches.
 EPEX SPOT. Marktdaten Day-Ahead-Auktion: Börse für Stromspotmärkte.
 Österreichisches Institut für Bautechnik. OIB-RICHTLINIE 6 - Energieeinsparung und Wärmeschutz(OIB-330.6-009/15).
 European Energy Exchange (EEX). European Emission Allowances | Global Environmental Exchange: Preis in EUR/tCO2.