BaMa Documentationbama.ift.tuwien.ac.at/static/downloadfiles/A1_Bama_Documentation.pdf · e!Mission.at - 1. Ausschreibung Klima- und Energiefonds des Bundes – Abwicklung durch die

2017

BaMa Documentation

DETAILED SOFTWARE DOCUMENTATION

BAMA KONSORTIUM

e!Mission.at - 1. Ausschreibung Klima- und Energiefonds des Bundes – Abwicklung durch die Österreichische Forschungsförderungsgesellschaft FFG

Page 1 of 70

Balanced Manufacturing (BaMa)

Authors:

Friedrich Bleicher (TU Wien- Institute for Production Engineering and Laser Technology)

Benjamin Mörzinger (TU Wien- Institute for Production Engineering and Laser Technology)

Christoph Loschan (TU Wien- Institute for Production Engineering and Laser Technology)

Ines Leobner (TU Wien- Institute of Energy Systems and Thermodynamics)

Peter Smolek (TU Wien- Institute of Energy Systems and Thermodynamics)

Wolfgang Kastner (TU Wien- Institute of Computer Engineering)

Bernhard Heinzl (TU Wien- Institute of Computer Engineering)

Iva Kovacic (TU Wien- Institute of Interdisciplinary Construction Process Management)

Georgios Gourlis (TU Wien- Institute of Interdisciplinary Construction Process Management)

Thomas Sobottka (TU Wien- Institute of Management Science)

Felix Kamhuber (TU Wien- Institute of Management Science)

Roman Klug (AutomationX GmbH)

Philipp Trummer (AutomationX GmbH)

Michael Mühlhauser (AutomationX GmbH)

Neat Likaj (AutomationX GmbH)

Gerhard Gotz (Berndorf Band GmbH)

Klemens Gregor Schulmeister (Berndorf Band GmbH)

Günther Daubner (Daubner Consulting GmbH)

Rudolf Zeman (Daubner Consulting GmbH)

Niki Popper (dwh GmbH)

Matthias Rössler (dwh GmbH)

Thomas Palatin (FHW Franz Haas Waffelmaschinen GmbH)

Erich Krall (GW St. Pölten Integrative Betriebe GmbH)

Josef Obiltschnig (Infineon Technologies Austria AG)

Karsten Buchholz (Infineon Technologies Austria AG)

Ewald Perwög (MPREIS Warenvertriebs GmbH)

Friedrich Koidl (MPREIS Warenvertriebs GmbH)

Gerhard Fritz (Metall- und Kunststoffwaren Erzeugungsgesellschaft m.b.H.)

Andreas Platzer (Siemens AG Österreich)

Kathrin Weigl (Siemens AG Österreich)

Albert Dechant (Wien Energie GmbH)


Page 2 of 70

Table of Contents

1 Documentation structure and guidelines .................................................................. 4

2 User level ................................................................................................................ 5

2.1 General description of the BaMa methodology and tool chain ................................. 5

2.1.1 Objectives and application areas of BaMa – What is it? .......................................... 5

2.1.2 Addressable problem types ..................................................................................... 6

2.1.3 Limitations to the application of BaMa ..................................................................... 7

2.1.4 Introduction to BaMa methodology – The charm of cubes ....................................... 8

2.1.5 Principle of the product footprint .............................................................................. 9

2.2 Defining the problem ............................................................................................. 10

2.2.1 Business case and questionnaire .......................................................................... 10

2.2.2 Optimization variables ........................................................................................... 12

2.2.3 Constraints ............................................................................................................ 13

2.2.4 Optimization goals ................................................................................................. 13

2.2.5 System boundary .................................................................................................. 14

2.2.6 Secondary benefits from the BaMa tool ................................................................. 14

2.3 Support process and implementation services ...................................................... 15

3 Conversion level – “Consultant” ............................................................................. 15

3.1 BaMa methodology ............................................................................................... 15

3.1.1 Fundamentals and definitions - BaMa Glossary .................................................... 15

3.1.2 Cube characteristics and interfaces ....................................................................... 16

3.1.3 Cube classes and sub-classes .............................................................................. 20

3.1.4 Cubing: the methodology to perform a cube segmentation .................................... 26

3.2 Product footprint methodology ............................................................................... 30

3.3 Technical description of required hardware and software ...................................... 31

3.4 Optimization .......................................................................................................... 32

3.4.1 Introduction ........................................................................................................... 32

3.4.2 Implementation and deployment of the first prototype ........................................... 33

3.4.3 Defining an objective function ................................................................................ 35

3.4.4 Phase 1 – selection of suitable meta-heuristics ..................................................... 35

3.4.5 Phase 2 – tuning and customization ...................................................................... 36

3.4.6 Implementation and deployment of the second prototype ...................................... 39

3.5 Change Process .................................................................................................... 41

3.5.1 Fundamentals ....................................................................................................... 41

3.5.2 BaMa Implementation Process in relation to EnMS ............................................... 42

3.5.3 Formulation of an energy policy (Item 4.3) ............................................................ 42

3.5.4 Conduct energy planning (Item 4.4) ...................................................................... 43


Page 3 of 70

3.5.5 Implementation of energy management system (Item 4.5) .................................... 45

3.5.6 Checking (Item 4.6) ............................................................................................... 46

3.5.7 Management Review (Item 4.7) ............................................................................. 46

4 Implementation level – “Provider” .......................................................................... 47

4.1 Requirement definition .......................................................................................... 47

4.1.1 Goals and Target Environment .............................................................................. 47

4.1.2 Product Use .......................................................................................................... 48

4.1.3 Product Overview .................................................................................................. 48

4.1.4 Product description ................................................................................................ 50

4.2 Implementation of the optimization framework ....................................................... 51

4.2.1 Implementation of the optimization framework - Vienna University of Technology . 51

4.2.2 Implementation of the optimization framework - AutomationX ............................... 60

4.3 System requirements and Hardware ..................................................................... 63

4.3.1 Measurement Structure ......................................................................................... 63


Page 4 of 70

1 Documentation structure and guidelines

The document at hand represents a target group-oriented communication strategy, containing all the

documented results of the project. The document will be gradually filled during the course of the project

as the contents are developed and will be published on the BaMa website. The intention of this document

is to adjust the level of detail to the needs of specific user target groups in order to increase the readability

of the BaMa specifications. Three main user groups were identified and their characteristics are presented

below.

BaMa users – User level:

This group represents the industrial users, who may be interested in the implementation of BaMa at their

plants. Within the project, this user group is represented by Infineon, MPreis, Haas, MKE, GW St. Pölten

and Berndorf Band. The main scope of the documentation for this user group is to give a brief overview of

the methodology and functionalities of BaMa and to provide a basis for the decision, whether BaMa

provides a feasible solution for the specific problem a user might want to address.

BaMa consultants – Conversion level:

This group consists of companies that offer consulting services and/or technical assistance for the

implementation of BaMa tools at factories. In many cases, the provider (see below) may also serve as the

consultant. This part of the documentation mainly focuses on the aspects that need to be taken into

account, when implementing BaMa into a specific factory.

BaMa providers – Implementation level:

The last target group consists of companies that want to develop BaMa tools. In the project, this user group

is represented by AutomationX and Siemens. In this part all the specifications, formulation and algorithms

necessary to do so are described. Of the three levels, this is by far the most detailed one.

The user levels differ in degree of detail from the “User” level as the least to the “Provider” level as the

most detailed one. However, knowledge of less detailed levels may be required for the more detailed ones.

Chapters and sections contained in this documentation are assigned to the corresponding deliverables,

as they are defined in the project’s handbook.


Page 5 of 70

2 User level

2.1 General description of the BaMa methodology and tool chain

2.1.1 Objectives and application areas of BaMa – What is it?

Balanced Manufacturing, short BaMa, tries to marry sustainability with competitiveness in industrial

production, by finding a balance between energy and cost efficiency.

More sophisticatedly expressed, it is a methodology and tool chain for monitoring, predicting and

optimizing the energy and resource demands of manufacturing companies under consideration of the

success factors time, costs and quality.

What sets BaMa apart from other energy management systems is its comprehensive approach. Although

a variety of methods and tools for optimization of energy efficiency exist, the majority focuses on singular

aspects of energy efficiency or application areas. BaMa considers more than one application field, when

analysing the system for optimization potentials. It addresses the production, the building, the energy

system and the logistics.

BaMa is not a design tool, which focuses mainly on finding improvement potentials in the design of a

production or the supporting infrastructure. BaMa is a planning tool that assists in operating a production

facility in a balanced, energy efficient way, much like an advanced planning and scheduling system would

support a cost and/or time efficient production process. Therefore, BaMa introduces energy as a steering

variable into the plant’s operational planning.

BaMa consists of two main parts: firstly, a method for a thorough analysis of the observed production

facility. For the method, a modular approach was chosen that segments the factory into "cubes", which will

be described into more detail below.

Secondly, based on the method, a tool chain realizes the monitoring, prediction and optimization. The

BaMa tool chain contains three core modules:

● Monitoring: Data on resource consumption are aggregated and visualized. Monitoring results serve

as a reporting document and help to identify technical and infrastructural optimization potentials. It

furthermore complies with the requirements of the energy management standard ISO 50001.

● Prediction: Based on data and numerical simulation models of machines, equipment and

infrastructure, this part of the tool chain allows forecasting of the overall energy demand of the

plant based on the production plan and other forecasting data (e.g. weather).

● Optimization: Using the results of repeated simulations, an optimization algorithm improves the

plant operation with regard to the optimization targets energy, time, costs and restrictions such as

resource availability and quality.

Based on monitoring data and simulation results BaMa derives a specific product footprint. The product

footprint represents a product’s expenditures concerning cost, time, energy and the environmental impact

such as resulting carbon emissions in the product life cycle phase within the factory (gate-to-gate). The

BaMa tool chain is generally embedded into the automation system of the company.


Page 6 of 70

In addition to the technical tools, BaMa provides guidelines for a change management process based on

the ISO 50001 to accompany the implementation.

2.1.2 Addressable problem types

As mentioned above BaMa is a planning tool that is designed to support the operation of a production

facility in a balanced, energy efficient way. Therefore, BaMa ideally is utilized periodically during the

operation of the plant in order to compute optimized strategies for the plant operation.

A typical case of application could be the following:

Production schedules for the upcoming week are compiled on Fridays. Therefore, BaMa will be run once

a week on Friday, calculating a production schedule taking into account deadlines, available resources

like machines or staff while trying to find a balanced optimum. This could be realized by avoiding

scheduling of energy intensive processes at the same time, in order to prevent cost intensive peak loads.

Waste heat inducing processes could substitute for room heating demands. Maybe savings from shifting

processes with high electricity demand to hours with cheaper supply outweigh increased personnel costs.

There are varieties of challenges that can be addressed with the help of BaMa. Figure 1 gives an overview

of conceivable results for the application fields and the different modules of BaMa in the four fields of

action.

Figure 1 Applications of BaMa


Page 7 of 70

It is not mandatory to address all application fields or implement all modules (except prediction is

mandatory for optimization). However, the more comprehensive the implementation the more likely a

global optimum will be achieved.

2.1.3 Limitations to the application of BaMa

Although BaMa can address many challenges faced with when dealing with energy efficiency, it has

limitations, which shall be addressed here.

Due to its nature as a planning tool, that is supposed to support the daily operation of a plant, BaMa does

not provide some functionalities:

● BaMa cannot optimize the technology of a production process itself (e.g. a NC program for a certain

production process). This requires specialized technological knowhow that a quite universally

applicable energy efficiency optimization is unable to provide.

● BaMa is no design tool. It does not offer automated optimization of design or dimensioning of

equipment, infrastructure or planning of layouts. It is strictly designed for optimization of the

operation of existing equipment.

● BaMa can only assess a product’s environmental impact during its production process (gate-to-

gate). Other parts of the product life cycle, such as transport beyond the gate, distribution, use are

not within the scope. BaMa is also not a tool for sustainable product design.

● BaMa is a planning tool, not an automatic control. This means it has to be operated by humans

and it suggests strategies that have to be realized by humans. However, BaMa may influence

automatic controls.

Other crucial points BaMa needs in order to operate successfully are degrees of freedom and a database:

● In order to optimize, BaMa needs degrees of freedom. If there is none, optimization is impossible.

For example, a company with one machine that is utilized 24/7 for the same process and only uses

electricity obtained directly from the supplier has no degrees of freedom.

● On the other hand, BaMa can only handle a certain number of degrees of freedom. If the system

becomes too large, computation time or required processing power will grow beyond a feasible

point.

● BaMa cannot perform an optimization without data about energy, costs and availability of

resources. This data will have to be collected from monitoring, measurements or other data sources

(i.e. ERP, MES). This means that BaMa in most cases cannot be a standalone system, but will

need an automation system with certain functionalities to rely on.

Finally, there are some restrictions concerning the type of production:

● The need for a reliable database implies that companies with often changing production processes

with energy demand profiles that are not derivable from general principles, are not an ideal

application field for BaMa.


Page 8 of 70

● BaMa is targeted towards optimization of the energy demand of an industrialized production. It

does not optimize the energy demand of administration or other office related business units. If a

production accounts for only a small part of the overall energy consumption or is mainly manually

executed, BaMa is not the right tool.

● For industries with large energy demands and very specific production-processes (e.g. steel

production, paper production...) specialized optimization tools or strategies exist. They may be a

better solution than BaMa.

2.1.4 Introduction to BaMa methodology – The charm of cubes

In preparation for the implementation of the BaMa tool chain a production plant must be analysed. This is

part of a methodological approach to address the high system complexity and heterogeneity by dividing

the overall system from an energetic point of view into well-defined manageable modules. Those then

allow a focused system analysis independent of the surrounding environment. The methodology BaMa

uses to do so is formulated at a very generic level to ensure its usability in a variety of production facilities.

The basic module of this system analysis is called “cube”. Cubes have a clearly defined interface and

represent a certain physical behaviour that contributes to the resource balance of the overall system.

Cubes are balances. This approach, well known from fluid mechanics or thermodynamics, was chosen

over the usual process approach, because it makes the system more conveniently manageable in

prediction and optimization. Cube boundaries bundle balances in terms of energy-, material-, cost- and

information flows all in the physically same place and therefore intersect the whole system into observable

parts.

Figure 2 Balances with and without cubes

The prediction and optimization modules use models of the cubes (virtual twins of the real cubes) in order

to simulate the behaviour of the system parts. The cube concept unfolds its charm here, because the

characteristics and attributes of cubes can be specified in a generic way, which guaranties the applicability

of simulation models for all parts of the plant and for different kinds of productions. Figure 3 shows the


Page 9 of 70

relationship between real and virtual cubes in the simulation environment and the integration into the

overall automation system architecture.

Cube boundaries are of course imaginary. However, in most cases they coincide with some sort of physical

representation. A cube can be for instance a machine tool, a chiller, a baking oven, the production hall or

a utility system. In a BaMa- monitoring system sensors would preferably be installed to measure flows at

cube interfaces.

Figure 3 Architecture of the BaMa tool chain, including real and virtual cubes

2.1.5 Principle of the product footprint

The main objective of BaMa is to define a significant product footprint, which assesses the impact of the

manufacturing of products on the environment and to describe a methodology for data aggregation. A

product footprint sets the product success factors in context with its ecological impact. The considered

expenditures are energy, cost, carbon emissions and time.

In general, for defining a footprint investments are aggregated over the entire life cycle of a product, from

the cradle to the grave. The life cycle can be divided into four phases starting from the harvesting of the

resource materials, to the manufacturing process and the product’s usage and finally ending with its

disposal and recycling. To be in line with the cube concept a bottom-up approach is the logical

consequence as it allows for real-time evaluation of a batch or even single product using monitoring or

simulation data. The classification of the different kind of footprints is shown in Figure 4. The product life

cycle footprint is the total footprint during the lifetime of the product. It can be classified by the four

components of the footprint and by the phase in the product life cycle.


Page 10 of 70

BaMa however is focusing on the product footprint during the production phase of a product’s life cycle.

This partial footprint using the gate-to-gate approach is referred to as the product-manufacturing footprint.

This allows for a detailed assessment of the manufacturing processes and a reliable calculation of the

footprint aggregated during production. The bottom-up approach contributes significantly in achieving the

necessary level of detail. Each process in a production plant is determined and modelled in a hierarchical

way. On the process level, balances regarding footprint calculations are also implemented. The result is a

framework, which integrates the connections between individual processes and captures the complexity

of the entire production plant. This allows attributing the product footprint, from gate-to-gate, separately to

each manufactured product.

Figure 4 Classification of product footprints

2.2 Defining the problem

2.2.1 Business case and questionnaire

A questionnaire has been designed in order to develop a use case and identify key issues and problems

of a production facility. The idea is to get a better understanding of the framework in which the BaMa

methodology can be implemented on a particular application-partner and clarify the benefits which the

“user” wants to achieve. Additionally, a structured way of defining the use case in writing helps in avoiding


Page 11 of 70

misunderstandings and balances the expectations of all sides involved. An overview of the business

questionnaire can be found in the following page.


Page 12 of 70

2.2.2 Optimization variables

Using the prediction of the behaviour of the plant the next step is to influence the projected scenario to

achieve better results. The aspects that are being varied need to be defined in order to achieve savings in

BaMa Questionnaire

Description of the Scenario

How will the company use the BaMa tool chain? List the different applications and describe them.

What degrees of freedom are available for the applications?

How does the company benefit from implementing the different applications and in total?

o Differentiate by economic, ecological and socio cultural benefits.

o Differentiate by direct and indirect benefits.

What modules of the BaMa tool chain will be implemented and at what scope?

What degree of acceptance for the implementation of the BaMa tool chain is to be expected? (short and

long term)

Specification

Which technological elements are needed?

o List the required Software tools.

o List the Hardware requirements.

o Describe the Integration of the BaMa tool chain into the company’s IT and automation

services.

o List the minimum requirements of company’s IT and automation services.

What measuring concept is implemented at the company? (e.g. mobile or fixed measurements)

o Present a rough preselection of required measurement points.

Is the BaMa tool chain going to be used temporarily (e.g. as part of an energy efficiency project) or

permanently (e.g. during normal operation)?

How frequently is the BaMa tool chain going to be used during normal operation?

What competences do users of the BaMa tool chain need?

Roadmap

Compile a rough time schedule of the implementation including milestones.

Compile a resource map for the required know-how and a plan to access the resources.

Business Case

What investments at the company need to be made?

o Differentiate between financial, human and technological resources.

o Differentiate between the implementation phase and the operation phase.

Estimate of the benefit potential

o Describe the economic benefits (revenue or savings per year).

o What other benefits can the BaMa tool chain provide to the company?

(e.g. product quality, customer satisfaction, image gain, strategic benefits,...)


Page 13 of 70

energy, cost, time or carbon emissions. In other words, the goal is to reduce the total footprint of the plant

within a given time period.

As mentioned in the business case questionnaire, for the optimization of the plant operation first the

potential degrees of freedom need to be found. Only being able to change aspects of the plant operation

allows reducing the total footprint. The degrees of freedom can range from a change to the production

schedule to varying process parameters or to the control of the energy systems. Once found, specific

variables that represent the degrees of freedom need to be defined, called optimization variables. Those

variables are the only parameters that can be modified during the optimization process. Using the

aforementioned examples, optimization variables can be the starting times of jobs on a production

machine, the set point temperature inside an oven or the control strategy of multiple chillers.

As a rule of thumb it can be mentioned, that a higher number of optimization variables yields higher

potential savings but also increases the system complexity. Depending on the case, a trade-off between

savings and performance has to be made.

2.2.3 Constraints

Constraints represent the limitations imposed on the optimization variables. Commonly the optimization

variables are not allowed to reach arbitrary values but are limited to a certain range. Two categories of

constraints exist: hard and soft constraints. Hard constraints are absolute and cannot be violated. Soft

constraints can be crossed but doing so results in a penalty, which makes it less desirable to do so. The

definitions of constraints reduce the number of possible outcomes of an optimization variable during the

optimization process, resulting in a smaller potential for savings. Therefore the implementation of

constraints should be avoided when possible and only setup when inevitable. Constraints can range from

legal restrictions, to resource availability or capacity limits.

2.2.4 Optimization goals

The goal of the optimization process is to reduce the total footprint of the plant regarding ecological and

economic factors, resulting in a balanced manufacturing. The components of the footprint are cost, energy,

carbon emission, time and quality. The quality of the products is highly important to the overall success of

a company and is not to be compromised. All of the steps in implementing the BaMa tool chain are aimed

at preserving the quality of the products. As a result, the success factor quality will not be considered

directly when optimizing the plant operation. However, a thorough analysis of the processes in the system

helps finding problems in manufacturing and can lead to improvements of quality as a secondary benefit.

The remaining optimization goals are cost, energy, carbon emission and time. These factors are simulated

with the BaMa-Prediction using a bottom-up approach, leading to a very detailed footprint of the plant. The

four components of the footprint represent the target function of the optimization and therefore the

optimization goal. Depending on the importance of the different aspects of the footprint, weights are used

to represent the priorities of the company. Figure 5 shows a possible weighing of the elements of the

optimization goal.


Page 14 of 70

Figure 5: Visualization of prioritizing optimization goals

Depending on the set priorities, the result of the optimization process can vary. The scenario representing

the least carbon emission is not necessarily the same as the least cost. Usually a balance that best suits

the overall needs of the company is chosen.

2.2.5 System boundary

The system boundary determines the investigation area of a use case and defines which systems,

processes and facilities are to be included within the BaMa study. It sets clear limits between the analysed

system and the surrounding environment as well as the conditions according to which interactions occur.

The selection of the system boundary should be consistent with the optimization goals of the use case.

Decisions shall be made regarding, which units of the provided infrastructure, in accordance to the four

BaMa classes (machines and production process, building, technical building services and energy

systems, and logistics) are to be included in the analysis and to which level of detail these units shall be

studied.

The selecting of cut-off criteria, used for the system boundary establishment, shall be identified, clearly

stated and explained where necessary. In case the system boundary refers to a partial procedure and

omits processes, facilities and sub-systems that in general interact with each other, it shall be clearly stated

as well as the reasons and criteria of such a practice shall be explained.

Generally, the BaMa methodology adapts a gate-to-gate approach while studying a manufacturing facility.

Focus is given to the production process and all entities to be included in the system boundary are direct

or indirect components of this procedure.

2.2.6 Secondary benefits from the BaMa tool

In addition to direct benefits of optimizing the plant operation secondary benefits occur. In general the

thorough analysis of the processes of production, logistics, energy system and the building, results in a

better understanding of the plant itself. Errors during installation or setup of equipment can be recognized


Page 15 of 70

or energy intensive processes identified. Using this information, an investment plan can incorporate the

replacement of equipment with unnecessary high energy demand. With the detailed information of the

energy consumption including costs, a return of investment can be derived with little effort.

Another necessity for implementing the BaMa tool chain is the examination of the material flow inside the

facility. Again, simply the analysis of the system can lead to the identification of problems. Extended

storage times between machines or long transport routes can be avoided. By studying, the way orders are

handled miscommunications can be reduced and the workflow tightened.

A more detailed explanation of secondary benefits follows in D3.2.

2.3 Support process and implementation services

Guidelines and support will be provided for the implementation of BaMa at industrial plants and facilities.

That contains assisting partners to launch the BaMa tool chain at their facility while also creating

awareness about energy efficiency related topics. Support services will also include on-site training

workshops and establishment of user roles, responsibilities and permissions. More details on the subject

regarding the successful implementation of BaMa tool chain in a company will be included in the

documentation of D7.1.

3 Conversion level – “Consultant”

3.1 BaMa methodology

3.1.1 Fundamentals and definitions - BaMa Glossary

BaMa tool chain

Collection of all software that implements the functionality of BaMa. The BaMa tool chain consists of the

BaMa-Monitoring, BaMa-Prediction and BaMa-Optimization.

BaMa-Framework

BaMa tool chain without use case specific virtual cube models. The BaMa-Framework is the part of the

BaMa tool chain that is independent of the specific application (and thus the same for every use case).

BaMa-Monitoring

Hard- and software employed in BaMa for collecting, processing and visualizing measurement data. The

BaMa-Monitoring software is part of the BaMa tool chain.

BaMa-Prediction

Part of the BaMa software used for prediction of relevant state values and other information (e.g. total

energy consumption, product footprint) by means of simulation.

BaMa-Optimization


Page 16 of 70

Part of the BaMa software used for determining optimal operational management strategies for the system

under consideration. The BaMa-Optimization makes use of simulation functionality implemented in the

BaMa-Prediction module.

Cube

Encapsulated part of the system under consideration. Cubes are part of a methodological approach for

modularization and hierarchical structuring of the overall system from an energetic point of view into well-

defined modules with manageable complexity, which then allow a focused system analysis independent

form the surrounding environment. Cubes consolidate all relevant resources (energy, materials, etc.) within

identical system boundaries and interact with each other exchanging energy, material and information flow

via uniformly defined interfaces (see “Cube Interface”). The conceptual modularization allows on the one

hand systematic monitoring by equipping the cube interfaces with measuring devices for detecting

incoming and outgoing energy and resource flows. On the other hand, it is also used for developing

simulation models (as part of the BaMa-Prediction software) where cubes have a “virtual counterpart” (so-

called virtual cube) in the form of a component in a simulation model.

Cube interface

Interface at the boundary of a cube, through which a cube can interact with other cubes by exchanging

energy, material or information. Real cubes as well as virtual cubes have interfaces:

- The interfaces between two real cubes are at the imaginary boundaries between them. These

interfaces can be equipped with measuring devices to detect incoming and outgoing flows.

- The interfaces between respective virtual cubes have to be implemented in the simulation model. All

energy, material or information exchange is represented as information flow in the form of continuous

variables (energy, information) or discrete entities (material).

Product footprint

Product-related measure for the total amount of resources (costs, energy, CO2 emissions and time)

necessary for the processing of individual products or batches of products within the system boundaries.

3.1.2 Cube characteristics and interfaces

The generic term "Cube" describes an encapsulated part of the observed overall system. This is part of a

methodological approach to address the high system complexity and heterogeneity by dividing the overall

system from an energetic point of view into well-defined manageable modules (Figure 5), which then allow

a focused system analysis independent of the surrounding environment. Integrating different viewpoints

and areas of engineering (machinery, energy system, building, and logistics) in a single system description

makes it necessary to establish a general specification of the cube properties and interfaces.


Page 17 of 70

Figure 5 Production facility as interacting cubes

Cubes consolidate all information and resource flows (energy, materials, etc.) within identical system

boundaries, which not only promotes transparency during simultaneous analysis of energy and material

flows, but the obtained modularity also increases flexibility for adaptation to specific environmental

conditions. Cubes have uniformly and consistently defined interfaces, through which they interact with

each other by exchanging energy, material and information flows, see Figure 6.

Figure 6 Generic cube interfaces with energy, material and information flows

The material flow incorporates the immediate value stream (e.g. work piece, baking goods) and is

described as discrete entities. All necessary energy flows (electrical, thermal, etc.) are represented as

continuous variables together with their respective CO2 rates and is quantified inside the cube boundaries

using balance equations. Information flow provides operating states and monitoring values for the higher-

level control as well as control actions for the cube module.


Page 18 of 70

This modular cube description and specified interfaces enables analysing and modelling of the internal

behaviour independent from its surroundings. For experimental analysis based on measurement data,

cube interfaces can be equipped with measuring devices to detect incoming and outgoing flows. The

modularization of the observed overall system is also used for developing simulation models for these

cubes. Thereby, cubes have a “virtual counterpart” (so-called virtual cube, see Figure 7) in the form of a

component in a simulation model. The retained encapsulation and interaction via defined interfaces

provides flexibility during internal modelling of the cubes (e.g. as mathematical ordinary differential

equation models, data models, etc.) and for reusing implemented components in other models.

Figure 7 Architecture of the BaMa tool chain including a virtual representation of the observed system

Figure 7 shows the relationship between real and virtual cubes in the simulation environment and the

integration into the overall automation system architecture. The BaMa tool chain obtains measurement

and status data from different levels of the automation system and on the other hand delivers prediction

data and proposals for optimized operation strategies that can be adopted - with user interaction - in the

real system.

Following an object-oriented approach, the generic interface and attributes definition of the cubes can then

be implemented in an abstract class, from which specialized cube models can be derived as hierarchically

arranged classes and subclasses, see Figure 8. These classes implement a specific inner cube model

structure depending on the category (machines and production process, building, energy system and

logistics), for which suitable modelling approaches have to be developed.


Page 19 of 70

Figure 8 Cube classes and sub-classes

To model a given use case of a production facility, these classes are then instantiated in concrete models,

parameterized and connected via the interfaces, therefore mapping the topology of the system under

consideration. Cube instances represent specific parts of the overall system and can in turn be composed

of other cube classes (composite cube), thereby enabling hierarchical structuring of the overall model

instances. For example, a composite cube may consist of a production cube instance with input queue

(logistics cube) and local compressed air supply (energy system cube), and be in turn itself located within

a thermal zone cube. The internal details of the composite cube are in many cases not relevant to the

outside, so it is sufficient to represent the composite cube as a black box on a higher level, thereby hiding

unnecessary model complexity and gaining clarity. In the same way, composite cubes also allow iterative

model development with subsequent refinement of specific sections depending on the necessary level of

detail without influencing the surrounding cube models.

On the downside, hierarchical modelling also comes with some restrictions. For one thing, it is extremely

important that all cube interfaces be defined uniformly as a composite cube must provide interfaces for all

its sub-cubes and subsequent changes to the interfaces would require significant effort. In addition, the

intuitive tree structure for cube instances limits nesting possibilities. Since a cube can only be sub-cube of

at most one composite cube, for example, production and logistics cubes spanning multiple thermal zones

have to be divided at the zone boundaries in order to nest them inside respective cubes (Figure 9).


Page 20 of 70

Figure 9 Example of cubes’ relations within a plant

The direct value stream is modelled in the virtual cube system as discrete entities. This is important for the

traceability of individual products within the system boundaries and for assigning them product-related

characteristics (e.g. product footprint). Entities as object instances have a unique identity and store

product-specific information in the form of attributes that define its properties. Some attributes are fixed

(e.g. ID), others can be modified along the way of the entity through the virtual system, for example to

aggregate energy and carbon emission amounts. Entities can be generated (e.g. waste from processing)

as well as collected and combined (e.g. batching, assembly).

3.1.3 Cube classes and sub-classes

Based on the interface definition and the generic attributes of the cubes, suitable modelling approaches

for all four cube classes were developed. A detailed description of the specific properties, functions and

input-output parameters of each cube class and its sub-classes follows in the four subsections bellow.

Machines and production process

The machine cube Figure 10 serves as a depiction of the smallest level of production processes.

Processes within the machine cube can be value adding or non-value-adding.

The input interfaces of the cube contain all recourses needed to conduct the production process. Input

information could contain knowledge about what part will be processed on what kind of machine in what

way, the raw material itself and the tools required for processing. Additionally the cube needs energy flows

(heat, cold, water, gas, etc.) to operate the machine.


Page 21 of 70

Figure 10 Schematic of the machine cube

The outputs of the production cube are separated in three different ways. First, there are the value adding

outputs, mainly the processed material (product) that moves within the value flow of the factory.

Furthermore, there is information that provides additional insight on the finished process. Second, there

are non-value-adding outputs that remain within the factory and will be used later during a production

process. Examples are unprocessed material, tools and facility material that are brought back into a

storage as well as industrial waste heat the company is using to heat a thermal zone. Finally yet

importantly, there are non-value-adding outputs that will leave the cube system and cannot be processed

further in a useful way. Besides the obvious waste material and chips there are broken tools, lost industrial

heat, cold, compressed air, electrical energy and lost information that is not collected and further

processed. All these aspects together describe the machine cube fully.

All processed parts within the cube need to have assigned values for costs, carbon footprint, and energy

calculations.

Building

The building construction cube (Figure 11) sets the spatial limit of its containing cubes and is composed

by and directly related to the thermal zone cubes. Thermal zone cubes host in turn other sub-cubes such

as machine cubes or energy system cubes. Changes at the building construction cube like renovations

and refurbishments or structural additions to the building envelope have a direct impact on the inputs and

outputs of the thermal zones and energy system cubes such as heating and cooling demand, level of

daylight, solar gains et cetera. Weather and the surrounding environment are regarded as information


Page 22 of 70

inputs on the building construction cube, but their results are allocated as outputs in the respective thermal

zones.

It is the highest cube class level in the hierarchy and therefore is used for handling inputs from value

stream entities reliant on sub-cubes concerning product footprint when necessary.

Figure 11 Schematic of the building construction cube

Each building construction cube consists of one or more clearly defined thermal zone cubes (Figure 12).

Although thermal zones, generally defined as air volumes at uniform temperatures bounded by heat

transfer surfaces1, are more a thermal rather than a geometric concept, the two are often connected by

linking building usage and operation with spatial layout and HVAC distribution. This approach describes

the content of thermal zone cubes. A zone could be a single room, a hall, an unconditioned attic or a

specific group of spaces in a building, for instance a hallway, four service rooms and a WC-core. It may

also be necessary to divide large spaces in multiple zones. In this case, it is important to use virtual

boundary objects, such as "air-walls" between adjacent zones within the same volume.

Thermal zone cubes interact with each other, or where appropriate with their outdoor environment, in

terms of transfer of state information and thermal flows. They require energy from different energy

system cubes to create a suitable indoor climate for their hosted usage. This demand is documented as

an output parameter. However, fuels, resources and electrical energy are not directly consumed by the

zone but from its serving HVAC systems. Inputs to a zone are divided into two groups, those that come

from outside the zone and those from inside. Affect from the environment and weather or neighbour

zones constitute external inputs while impacts from sub-cubes, value stream entities or employees are

internal inputs.

1 DOE 2013, U S Department of Energy, Getting started with EnergyPlus; Basic Concepts Manual – Essential

Information You Need about Running EnergyPlus (and a start at building simulation) University of California,

California, USA.


Page 23 of 70

Figure 12 Schematic of the thermal zone cube

Energy system - Technical building services (TBS)

To model the cubes that represent the energy systems, the inputs and outputs necessary for the

description have to be identified. In comparison to the machine cube, no value flow passes through,

meaning that the product does not get in contact with the cube. Therefore, the inputs and outputs are

composed solely of energy and information flows.

The purpose of the energy systems cubes can be categorized into two sub-classes: Energy converters

and energy networks, with the latter including energy storage.

Energy converters transform energy from one type to another, having multiple educts and products. An

example is an absorption chiller, which converts heat and electricity to cold and waste heat.

Energy networks distribute energy from inputs to outputs. Again, multiple inputs and outputs are possible.

The storage capacity of the network, as well as designated storages can be taken into account. As an

example, a warm water network with a water tank can be considered.

The energy flows carry a footprint factor to assess the CO2 emission and the cost of the energy used. In

comparison to the production cubes, it is not possible to add the footprint to a product directly since no

product passes energy system cubes. Therefore, the new footprint factor is calculated and passed on to

the output energy.

Energy Converter Cubes

Figure 13 shows the model of the energy converter cube. The full arrows represent the energy flow as

needed for the conversion. The dashed lines show information flows that communicate the demand for

the primary product as an input and the demand for resource energy as an output. The unusable waste

heat is an output into the thermal zone.


Page 24 of 70

Figure 13 Schematic of the energy converter cube

Energy Network and Storage Cubes

The second sub-class is the energy network and storage cube, as can be seen in Figure 14. The energy

sources feeding the network are inputs into the cube. The energy sinks, to which the energy is distributed,

is represented through the energy output. The sinks send a demand for the required energy as an input

and in return, the demand for the controllable energy sources is an output. Non-controllable sources deliver

into the network but no energy can be demanded since the source of the energy cannot be controlled. The

storage potential of the network or a designated storage is implemented by the differential energy change

inside the cube.

Figure 14 Schematic of the energy network and storage cube

Logistics

The logistics cube consists of cubes for transport and handling systems on the one hand and storage

system cubes on the other hand. The main characteristics are as follows:

Transport Cubes:

● change the position of production material in the course of a time interval

● they are the major elements in facilitating the material flow within the production facility


Page 25 of 70

● two main variants: continuous conveyors (i.e. conveyor belts) and discontinuous conveyors (i.e.

forklifts)

● determining material flow characteristics of the transport cubes are

o their stacking behaviour

o the ability to allow/disallow the sequence of the transported goods (i.e. the production

material)

o the routing logic for the transported goods

Handling Cubes:

● can change the orientation of production material and are able to reposition the material – i.e. from

a palette into the input buffer of a machine

● can sort production material in a defined mode

Storage Cubes:

● retain production material for a certain amount of time

● can provide defined conditions for the stored goods (e.g. temperature, humidity)

● have a defined mode of forwarding goods and changing/retaining the sequence of the stored goods

(e.g. first in first out, last in first out, time or priority based rules, …)

● common variants: incoming goods storage, finished goods storage, intermediate storage, buffer

storage

● in the BaMa context the storage cubes also comprise the transport and handling systems within a

storage area, e.g. automatic rack feeders and other shelf access equipment

The logistics cube shares most characteristics with the machine cube, as the material flow, e.g.

represented by C (Figure 15), enters the cube at the beginning and leaves this cube after a certain storage

or transport period. The difference between logistics and machine cubes is the lack of major processing

steps being conducted in the former and the lack of sorting and changes in position of production material

in the latter. In the majority of cases, these logistics cubes do not add value to the product. Sometimes

however value is added, for example when products have to rest a specified amount of time on the

transport or storage system – e.g. in the case of cooling on conveyor belts – or when handling steps are

necessary to perform production processes. Thus, it is important to note that these two cube categories

do not qualify as proper sets. Some machine cubes may perform sorting and transport functions to a

certain degree while some logistics cubes perform necessary processes and add value to production

material. They are fuzzy sets, production equipment should be categorized into machine, and logistics

cubes according to the predominant purpose of the respective piece of equipment.

The logistics cube has a specific demand of primary and additional energy resources and thus a defined

amount of energy consumption. The demands A and B (depicted in the model description – Figure 15)

serve as output for the communication with other cubes, while the consumptions A and B are defined as

input values. As within the borders of a machine cube, all passing parts need to have assigned values for

cost, carbon footprint and energy calculations that are adequately added to the products leaving the cube.


Page 26 of 70

Figure 15 Schematic of the logistics cube

3.1.4 Cubing: the methodology to perform a cube segmentation

In order to make cubing (=intersecting a production facility into appropriate cubes) more comprehensible,

this chapter will try to illustrate the process by showing some examples and compiling a list of questions

that should be resumed when sectioning a system into cubes.

The production facility of the example is a bakery2. It is housed in a one-storey building, the layout of which

is shown in Figure 16.

Figure 16 Layout of example bakery

2 Please keep in mind that this example is constructed to showcase the cube methodology and may not be entirely accurate or reasonable from

a technological point of view.


Page 27 of 70

Production and logistics

This bakery produces rolls and different sorts of bread and for this purpose owns two production lines, one

for the rolls and one for all kinds of bread. The raw materials are delivered on the left and stored in the

goods receipt storage. The doughs are mixed in an automated mixing unit. From there, the dough

containers are transported manually to the roll production line or the bread production line, where they

may queue before entering the production. On the roll production line, the dough is automatically portioned,

formed and inserted into the baking oven straight afterwards. The dough for the bread is portioned and

formed. Afterwards the prepared loafs are fermented in fermentation cells. The duration of the fermentation

may vary according to the kind of bread. In addition, it may be influenced actively by altering the

temperature in the cells. Afterwards, the bread is baked. The finished goods, rolls and bread are

transported to the packaging in engine-driven transport carts. After the packaging, the finished products

await the dispatch in the outgoing goods storage. Transports in the outgoing goods storage are again

performed manually.

Building and energy system

The production area in the building has three main zones. The goods receipt storage, the production zone

itself and the packaging and dispatch zone. Those three zones are conditioned differently. The incoming

goods storage is held at 18°C in order to ensure maximal shelf life of the raw materials. The production

zone is held between 20° to 24°C and is well ventilated to provide agreeable working conditions for the

staff and the packaging zone has to maintain a temperature of 12°C according to legal regulations.

Furthermore, there is an unconditioned TBS centre and an adjacent office conditioned to 20° to 24°C.

The heat for room conditioning is provided by a gas heater, fired with natural gas and the cold is provided

by an electricity-powered chiller. The two baking ovens are heated with thermal oil, which is heated by

gas. This energy system requires five grids. Cold and heat grids for room conditioning, a gas grid, a thermal

oil grid that also has a buffer heat storage and an electricity grid, which supplies all the production

machines, logistics equipment and infrastructure.

The given exemplary production will now be intersected into cubes. Please be aware that the given solution

may not be the only one but in this case seems to be the most feasible one.

Building cubes

The whole building has three different temperature set points within the building (12°, 18° and 20° - 24°C)

and one unconditioned area, which suggests at least four different thermal zone cubes. The two areas

with a temperature range from 20° - 24°C, however, differ in their thermal behaviour as far as that one of

them is ventilated, the other one is not, which makes a division into two thermal zones necessary. In total,

six building cubes where developed.

● Building construction cube

● Incoming goods storage thermal zone cube (18°C)

● Production thermal zone cube (20° - 24°C, ventilated)

● Packaging and dispatch thermal zone cube (12°C)


Page 28 of 70

● Office thermal zone cube (20° - 24°C, ventilated)

● TBS centre thermal zone cube (unconditioned)

Please note that the boundaries between the thermal zones must be exactly specified concerning their

geographic position. In most cases, it makes more sense to locate them on either side of the wall, rather

than in the middle.

Energy System cubes

Cubing the energy system is rather simple, according to chapter 3.1.3.3 and storages are in most cases

adequate cubes. Therefore, the energy system consists of eight cubes:

● Thermal oil heater cube

● Chiller cube

● Gas heater cube

● Gas grid cube

● Thermal oil grid and storage cube

● Electric grid cube

● Cold grid cube

● Heat grid cube

Storages and grids are in most cases both modelled as energy balances. In these cases, having both

inside one cube is perfectly fine. If the network needs to be modelled in more detail, (e.g. hydraulic

phenomena are taken into account) it may be necessary to separate them.

Machine and production cubes

In this example, the following six production cubes are defined:

● Mixing unit cube

● Roll production line and roll baking oven cube

● Bread forming cube

● Fermentation cube

● Bread baking oven cube

● Packaging cube

At the first glance, it might seem confusing that the roll production is one cube all together whereas the

bread line was sectioned into three cubes. The reason for this is quite simple. The rolls run through the

process on a conveyor belt without interruption or buffering. Once the process is started, the rest of the

line will invariably have to follow and process the rolls until they are baked. Therefore, the only temporal

degree of freedom of the whole line is the starting time. Whereas the duration of the fermentation and

possibly queue times between forming and fermenting can be subject of influence. Therefore, it makes

sense to section the bread production into more cubes.


Page 29 of 70

Logistics cubes

For the logistics cubes, all transport and storage utilities that actually require resources and therefore

change the value of the product are considered. There are four cubes within the bakery:

● Incoming goods storage cube

● Transport 1 cube

● Transport 2 cube

● Outgoing goods storage

The transport from the mixing unit to the production lines was not mapped as a cube because it is manual

and very short so the expended resources are very little and can be neglected. Furthermore the queue

before entering the production line was not considered a cube, since it does not require resources either

(the dough is just standing around) and can easily be modelled within the machine cube. The transports

from the production lines to the packaging however have to find a representation in the cubes since the

transport carts are motor-driven and therefore consume energy. The transport also had to be divided into

two cubes because it takes place in two thermal zones and interacts with both. A transport cube stretching

over both zones would violate the requirement of hierarchy.

Finally, the cubing delivers the result depicted in Figure 17 (exempting the grid cubes).

Figure 17 Cubes of the example bakery (except grid cubes)

After specifying the cubes, the interfaces between the cubes must be identified. The result is shown in

Figure 17. The transport cubes must be interfaced with the production cubes and among each other (see

changes in value stream in transport cubes 1 and 2). Apart from the energy flows in the grids, waste heat

is emitted from equipment and machinery cubes into thermal zone cubes, exchanged between thermal

zone cubes and with the surrounding. Furthermore, please note that the grid cubes do not exchange waste

heat with the thermal zones or the surrounding. This aspect was neglected due to efficient insulation of

the pipes. This missing interaction is also the reason, why the grid cubes do not have to be sectioned into

individual cubes according to the thermal zones they are in, as the transport cube had to be.


Page 30 of 70

Figure 18 Cubes and interfaces of example bakery

3.2 Product footprint methodology

Each part of the plant contributes to the product’s energy, cost or time consumption, as well as carbon

emissions, which accumulates the product footprint. The energy used by production machines, auxiliary

infrastructure, logistics and the building is aggregated from the entry of the raw materials to the departure

of the finished good. Therefore, the footprint methodology follows a gate-to-gate approach. For a full

product life cycle analysis the obtaining of the raw materials, the use-phase and the disposal of the product

have to be evaluated as well.

The method for aggregating the product footprint builds on the principle, that material and energy flows

can carry a footprint. This footprint is considered a conserved quantity, meaning that no footprint can be

dissipated or generated inside the system boundary. Additionally, only valuable flows can carry a footprint,

with valuable being defined as useful at a later stage. For example, the electricity produced by a generator

can carry a footprint, since it can be used for work or sold. The unusable waste heat of the generator due

to internal friction is not usable, therefore free of a footprint. Taking the system boundary of an entire plant,

the only output flows that can carry a footprint are products (material flow) or produced energy (energy

flow). Waste with resale value is considered a product and therefore has a footprint. The input flows carry

the footprint needed for extraction, processing and transport of raw materials or energy.

In each cube the balance of the footprint is calculated, where the footprint of the output flows is determined

by the footprint of the input flows. Different cube-classes have special challenges in aggregating the

footprint, however these general relations hold. A production cube performs a value-adding operation, in


Page 31 of 70

which a product is manipulated. Therefore, the footprint of the input flows of a production cube is

aggregated and attributed to the product. Since material flows are discrete, the footprint of the energy

flows are integrated over the time the product spends inside the production cube. Logistics cubes are

similar to production cubes. Although generally no value-adding operation is performed, the energy

necessary to transport, handle or store a product is added to the product footprint in the same manner.

Energy system cubes are not in contact with a product, therefore the footprint balance calculates the

footprint of the output energy flows. Since this is not a discrete quantity, no integration is necessary. Finally,

the building cubes calculate the footprint due to heating, cooling, air-conditioning and lighting and

aggregate it to all the products in a thermal zone at any given time.

From this bottom-up approach, different challenges arise. For example, the incorporation of standby-,

setup- and ramp-up times, the energy consumption of the administration and the allocation of different

products and by-products manufactured at a machine are some of the problems. The necessity to calculate

mean values and dividing them between different products demands for a way to assess the degree of

which each product is responsible for the generated footprint. The detailed documentation of the

Methodology of the Product Footprint explains the approach. It can be found at the BaMa homepage

[http://bama.ift.tuwien.ac.at/]. Furthermore, a detailed explanation for the definition of the data aggregation

for the product footprint will be part of D4.1.

3.3 Technical description of required hardware and software

The hardware requirements depend on the implementation. In the most minimalistic case, (the simulation

runs without any visualization or optimization) only a standard computer is necessary. The size of used

Random Access Memory depends on the size of the simulated model. For example, a simulation of a line

with 15 cubes takes approximately 150 Megabytes of RAM memory. There are no restrictions concerning

the CPU power, with state of the art computers. More CPU power, especially more cores, can reduce the

time needed for the simulation by a lot. The simulation is multi thread-able, so the required time will

decrease in an almost linear fashion to the amount of CPU cores. The software requirement for the

simulation is an operating system of Windows 7 or higher.

For the full installation of the BaMa-Tool (includes the simulation, optimization and visualization) the

minimum hardware specifications are at least a virtual machine with the following system metrics.

● 20 Gigabyte hard disk drive

● 4 Gigabyte of random access memory

● 1 processor core

These are the minimum hardware requirements for the BaMa-Tool. To reduce the processing time, it is

recommended to increase the RAM memory to 8 Gigabyte and raise the amount of processor cores to 4.

If many simulation data should be kept, the hard disk drive should be increased to 80 Gigabyte. As there

are more components for the full BaMa-Tool, additional software is also required. For the visualization of

the results and the configuration of the simulated orders, including the product data and the optimization

parameters, the software aX5 from automationX is required. This software includes the necessary PLC


Page 32 of 70

and visualization tools for the BaMa-Tool. Furthermore, Microsoft SQL database is needed for the storage

of the simulated data. The automationX software provides the visualization of the simulated data over an

interface to the database. The following list will summarize the required software for the whole

implementation of the BaMa-Tool:

● Operating system (Windows 7, Windows 10, Windows Server 2008 R2 or higher); the 64-bit

Version of the respective operating system is essential

● .net Framework 4.6.2 for the actual operating system

● automationX aX5

● Microsoft SQL Database (Express is possible)

The minimum required hardware specification for the aX5 software is:

● CPU: >= 2.0 GHz, 4 cores (reserved)

● Memory: >= 8 GB (recommended 16 GB)

● Hard disk Space: Operation System 60 GB

The visualization can be operated on a client computer, which is connected via network with the server. In

this case, the client computer has the following specifications for the hardware:

● CPU: >= 2.0 GHz with 2 Cores

● Memory: >= 4 GB (recommended 6 GB)

● Free harddisk space: minimum 5 GB

The software specifications do not differ from the server software specifications. The basic operating

system provided by AutomationX is Windows 10. The minimum installation of the BaMa-Tool is running

as a console application only. There are no other software packages needed, the results are saved in a

text document and the reflection of these saved results is not easy to handle. Furthermore, the console

application is only a prototype of the BaMa-Tool and should be fully integrated in the AutomationX aX5

Package in the future.

In the figure above the schematic structure of the full BaMa-Tool implementation is shown. The orange

box, which contains the simulation, model and cubes is the above mentioned minimum installation of

BaMa-Tool. This is part of the full BaMa-Tool which is shown in the green box and includes two more

implementations. The production plan generator creates the production plans which are simulated and the

optimizer rates the different simulations.

3.4 Optimization

3.4.1 Introduction

The BaMa Tool uses a simulation-based optimization to generate optimized production plans. In this setup,

the simulation serves as an evaluation function for the optimization: Initial plans are feed as parameters

into the simulation to determine the plan’s performance. The simulation in turn provides a feedback for

every plan (=intermediate solution), which is evaluated by an objective function. Based on the multi-criteria


Page 33 of 70

performance evaluation, the optimization algorithm creates new intermediate solutions to evaluate. This

process is repeated until a termination criterion is met – e.g. no further improvement of a plan or reaching

a time limit. Finally, the optimized plan is provided to the company IT and the planners to be implemented

in real life. Figure 19 gives an overview of the optimization structure.

Figure 19 BaMa optimization control loop

In order to fulfil an optimization the first steps include the setup of a well-defined simulation model for the

given use-case of an industrial bakery. This setup includes process sheets for the products, initial

production plans serving as initial solution and scenario-specific demand plans. In addition, the granularity

of the simulation has to be well thought out to receive the desired results. Finally, the target function has

to be set-up as well as an optimization algorithm. Another addition constitute the integration of hard and

soft constraints into the simulation model to reduce the fitness landscape describing the amount of

solutions according to a specific quality (“fitness”). The used optimization algorithm tries to search

efficiently through this fitness landscape. Concerning the setup of those topics, please see 51 4.2 on this

document. The following chapters deal with the implementation of two prototypes for one specific use-

case within the project BaMa.

3.4.2 Implementation and deployment of the first prototype

The optimization was developed using a test case implemented within the hybrid simulation developed in

this research project. The test case is a simplified model of the actual production facility in an industrial

bakery that will be implemented once the optimization module has been finalized. The simplification is

designed to offer a controlled setting, where the optimization results can be checked and validated

manually, while still providing the level of model complexity that the eventual use-cases in companies will

feature. It comprises a production line with two process variants for the production of baked and deep-

frozen rolls. The simulation model, displayed in the Figure below, features five major production machines

and four conveyor belts with deflectors/junctions and storage components in the production and logistics

system.


Page 34 of 70

Figure 20 Simulation model: prototype structure including material and energy flows

Two of the machines – an industrial oven and a freezer – feature a distinct thermal-physical behaviour.

Concerning the 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. 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 simulation was originally implemented in a Matlab®

framework and has then been implemented (C++ based) and improved by a software implementation

partner within the research project, thus providing the optimization development with a high-performance

– concerning simulation runtime – version of the hybrid simulation.

The actuating variables for the optimization comprise the order release times, the operating times and

modes for production equipment, including the machines, equipment in the production periphery and TBS

components. This enables the optimization i.e. to pre-heat an oven, to determine an energy efficient order

sequence and an optimal scheduling of orders with known delivery dates. Starting with an initial (set of)

solution(s), the optimization uses the hybrid simulation to evaluate the performance of every solution,

working towards finding a solution that optimizes a multi-criteria objective function.

Since none of the existing simulation based optimization approaches is directly applicable for this project,

a comparative evaluation of potential metaheuristics is being conducted. The procedure consists of three

phases:

● First, different metaheuristics that are suitable for complex multimodal (featuring local optima)

fitness-landscapes are being tested with identical scenarios and with a default set of optimization

parameters – we use the heuristics implemented in the Matlab® Global Optimization Toolbox.


Page 35 of 70

● Second, the best performing heuristics from phase one are then enhanced and customized to

provide an optimal fit for the given optimization task and model behavior and thus increase the

optimization performance – again mainly concerning processing time, with processing time being

one of the main factors for a practical application of the planning tool at the end of the project.

● Third, once finalized, the developed optimization module will also be implemented in the software

implementation, alongside with the already implemented simulation, for an improved tool

performance.

The metaheuristics compared in phase 1 are:

● Genetic Algorithm (GA)

● Particle Swarm Optimization (PSO)

● Simulated Annealing (SA)

● Pattern Search (PS)

● Multi Start Search (MSS)

3.4.3 Defining an objective function

The optimization evaluates the system performance for each parameter set (each plan/solution) according

to an objective function representing the management preferences of the planners. Through customizable

weights the multi-criteria objective function both normalizes different feedback parameters – i.e. costs,

delays and amounts of energy –, so that they are on a comparable scale, and prioritizes the part goals.

For the optimization development, a simplified version of the objective function is used, defined by the

industrial bakery that is the first application partner of the planning tool within the project. The function is

defined as a minimization function

The function 𝑓1 calculates storage costs for goods finalized before delivery date and penalty costs for

delayed deliveries as a function of the order completion time 𝑐𝑖. The other part-goals accumulate lot lead

times, overall energy usage multiplicated with their deriving costs and evaluate the number of unfinished

goods (with high penalties). The last part-goal evaluates the amount of power utilization above a defined

maximum power consumption threshold multiplied with the accumulated time of peak energy usage.

3.4.4 Phase 1 – selection of suitable meta-heuristics

The following (population-based and metaheuristics have been evaluated in order to find the most

suitable algorithm for the production-planning problem within BaMa:

● Genetic Algorithm (GA)

● Particle Swarm Optimization (PSO)


Page 36 of 70

● Simulated Annealing (SA)

● Pattern Search (PS)

● Multi Start Search (MS)

The GA shows the best overall behaviour: the results for multiple runs for the same scenario are consistent

and reliable; the relative goal-function improvement was between the two best performing algorithms in

the panel. Its robustness in large search spaces with multiple local optima make it a good fit for the

optimization task plus, it lends itself well to a targeted tuning and customization within the search algorithm

itself. Another advantage is the possibility to parallelize the simulation evaluations of one generation of

solutions – with the simulation runtime being the biggest time optimization-runtime component; this has

the potential to increase the optimization performance of the planning tool. This also concurs with the

findings of [28], who used the GA for a similar task of a simulation-based optimization in the field of

production planning, although for a less complex simulation model.

3.4.5 Phase 2 – tuning and customization

Having selected the GA for phase 2 of the optimization design process, we would like to give an overview

of the basic algorithm first (Figure 21). The GA mimics the process of evolution in biology: an initial

population (of solutions) is transferred to subsequent generations amid recombining and mutating

characteristics of the parents and selecting the members of the subsequent generation, according to their

fitness, as measured with the objective function [29].

Figure 21 Standard Genetic Algorithm


Page 37 of 70

One of the major problems the optimization metaheuristics faced in phase one, was the huge search space

and the large proportion of bad or unfeasible solutions therein. The search strategy of most algorithms

aims for a gradual improvement. If the vast majority of solutions is bad – i.e. not all products are being

produced –, the subsequent punishment through the objective functions for practically infeasible solutions

overshadows the feedback for gradual improvements. An obvious way to reduce the search space and

especially to reduce the space with bad solutions is to define constraints using specific knowledge of the

model behaviour. Still, those constraints must have an algorithmic character so that they can be integrated

into the optimization module generally – the following restrictions proved to be useful:

● preventing overlapping lot time-spans on any machine

● time windows for operating times of machines ≥ smallest lot processing time for that machine

● for every machine: sum of all operating time windows ≥ minimal accumulated processing time of

all production lots

These restrictions are based on the principle of setting multiple windows of operating time – each

demarcated by start and stop times – for every machine, depending on the number of production lots to

be produced. In addition to sensibly reducing the search space through constraints, tuning and customizing

the GA itself was the second major tuning component – this comprised:

● implementing a guided search by adapting the operators of the GA

● integrating a memory function from the Tabu Search algorithm

● implementing a mixed integer optimization

● hybridization: combining the GA with the PS

● determining the optimal population size

A guided search is meant to focus the parameter changes of the optimization algorithm on the most

promising changes. This, again, requires heuristic knowledge of the simulation model behaviour – in this

case, we again used the operating time windows already introduced for setting restrictions to reduce the

size of the search space. The guided search was implemented via the crossover operator of the GA as

follows:

● the original function of the crossover operator is to combine characteristics of two individuals of the

parent generation in order to generate one child with new characteristics – for the optimization

problem in this research and the corresponding vectors, this would lead to a large number of

practically unsound solutions, thus the function was altered to only feature a more sophisticated

mutation, which actually performs the guided search

● changing the production lot (order) sequence by swapping elements of the vector determining the

order release times

● shifting the operation time windows for each machine by shifting pairs of start/stop times within the

vectors containing the machine operating times

● contracting the operation time windows of the machines by a point mutation of start/stop times

(delay start & bring forward stop)

● changing the order release times by randomly mutating elements of the order release time vector


Page 38 of 70

● gradually decreasing the intensity of the stochastic changes described above – changing between

one and four vector elements at the start and only one towards the end of the optimization run,

while generally restricting all changes of time windows to below 25% – this is meant to explore a

wider variety of solutions (“global search”) at the beginning and conduct a more refined search

(“local search”) towards the end

Integrating a memory function was another means of reducing the optimization runtime, by preventing the

optimization from re-calculating solutions that the algorithm has already evaluated once before. This

function is an integral part of the Tabu Search algorithm, an optimization heuristic in its own right. In this

case, we have integrated the memory function within the GA based optimization procedure.

The mixed integer optimization is a means of further reducing the search space by restricting the step size

for the abovementioned vector changes. In a series of test runs with modulated step size, ranging between

1 and 3600 seconds, steps of 60 seconds proved to be the optimal step size for the given test case. For

this test case, the effect of restricting the step size was still only moderate but we expect it to be more

important in larger models for more complex real-life factories. The same is true for hybridization, by

combining the GA with a subsequent PS: The improvements in this test case have been minor – it remains

to be seen, whether this positive effect grows with an increasing size of the simulation model.

Figure 22 performance comparison (left side) and part goals development (right side)

Since the goal function contains multiple criteria, it was important to observe, whether the optimization was

able to work with and improve all part-goals. The above-standing figure shows the part goals trend during

the optimization run. The major development is the quickly lowered delayed delivery value (big weight),

the slow but consistent reduction of the energy value and the changing balance between the storage costs

and the lead times per production lot. The unfinished products value is kept at zero at all times due to its

big weight in the objective function.

The last major tuning step for the GA are tests with modulated population sizes. With multiple solutions

simultaneously being optimized, the GA is especially robust against being stuck in local optima and

stochastic fluctuations – as seen with the results of PS. However, multiple solutions in one generation also

means more evaluations and potentially an increased runtime of the optimization. Therefore, it is advisable

to determine the optimal population size. Thus, the population size in the experiments presented thus far


Page 39 of 70

was set at 1/8th x nvars (with nvars representing the number of variables in an optimization scenario). The

experimental results showed an increase in optimization speed when lowering the population size until

1/16th – beyond that point; the stochastic fluctuations increased considerably making the results

unreliable.

3.4.6 Implementation and deployment of the second prototype

The second prototype represents an enhancement of the first prototype, consisting of a more complex

production system (represented by cubes) including the consideration of technical building services.

Additionally results of a comprehensive measurement campaign have been added within the particular

cube classes. The production system consists of nine major production machines, nine conveyor belts

with junctions and two storage components within. 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 Figure 23.

Figure 23 Production system including production line, energy system and thermal zones

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 three cold networks and one electric network. The cold networks contain five chillers.


Page 40 of 70

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 parameterized 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 [9] 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:

The objective function is normalized to calculate a cost value denominated in Euro. The part-function 𝑓2

calculates storage costs for goods finalized before the delivery date while f1 calculates penalty costs for

delayed deliveries as a function of the order completion time. The next 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) [30]. 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 [31], evaluated with Spot-market prices of the European Energy Exchange AG [32] 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.


Page 41 of 70

3.5 Change Process

3.5.1 Fundamentals

The BaMa method is meant to be compatible with and facilitating the implementation of ISO 50001. The

ISO 50001 Standard descries the implementation of an energy management system, with the goal of

constantly improving the energy efficiency of companies by facilitating a continuous improvement process.

Throughout this document, the elements of the chance management process described in ISO 50001 will

be analysed concerning their connection to the BaMa toolchain and the application of the BaMa method.

The goal is to show how the BaMa method and toolchain can be used in the context of an EnMS and to

analyse the compatibility of implementing BaMa with ISO 50001.

The ISO 50001 Standard states…

It requires the organisation to establish, document, implement and improve its EnMS according to ISO

50001 standard. The organisation should define and document the scope and boundaries of its EnMS as

well as how to achieve continual improvement of its energy performance and EnMS. (Item 4.1)

The purpose of ISO 50001 standard is to enable organisations to establish systems and processes

necessary to improve energy performance. The standard applies to all factors affecting energy use that

can be monitored and influenced by an organisation. ISO 50001 standard does not specify energy

performance criteria. It provides a general-purpose system that allows organisations to choose

performance standards that they deem best meet their requirements.

Prior to developing the EnMS, the organisation should define the scope and boundaries of its management

system. The scope refers to the extent of activities, facilities and decisions that the organisation addresses

through an EnMS, which can include several boundaries. The boundaries are defined as physical or site

limits and / or organisational limits as defined by the organisation that could be a process, a group of

processes, a site, an entire organisation and multiple sites under the control of an organisation.

The focus of an ISO 50001 EnMS is on improving management processes, practices, and procedures that

control an organisation’s functions and activities with significant energy use. The overarching intent is that

by implementing a management process and continually improving this management system, it will

eventually lead to an improved energy performance.

In order to implement an energy management system according to ISO 50001, a defined process has to

be executed. The implementation of the BAMA Toolchain can be implemented with this process and its

results can contribute to achieving the results expected from an energy management system.

Consequently, in the following document, a reference change management process for the implementation

of an energy management system facilitating the tools provided by the BAMA toolchain is described.


Page 42 of 70

3.5.2 BaMa Implementation Process in relation to EnMS

Figure 24 Energy management process according to ISO 50001


It requires the top management to demonstrate its commitment to support and continually improve the

effectiveness of the EnMS. (Item 4.2.1)

It requires the top management to appoint a management representative(s) to promote awareness and

oversee the implementation of the EnMS. (Item 4.2.2)

Although not directly reflected in the elements of the BaMa implementation process, this step of ISO 50001

is the basis for the successful implementation of any further endeavours, including the implementation

itself. Not just does the commitment have to come from the top management of a given company, it also

has to go hand in hand with the assignment of the necessary monetary, personal and technical resources

to establish and uphold a successful energy management system.

To ensure effective operation of the EnMS, top management is required to appoint a management

representative and approve the formation of an energy management team. The management

representative (MR) is responsible for managing all aspects of the EnMS as it evolves. MR should have

sufficient authority, competency and resources to ensure the overall effectiveness of the EnMS. The

energy management team is responsible for ensuring the implementation of actions / measures of the

energy management decisions. The composition and size of the energy management team should be

determined with due consideration of the size and complexity of the organisation.

3.5.3 Formulation of an energy policy (Item 4.3)


It requires the organisation to define an energy policy to state its commitment for achieving energy

performance improvement. (Item 4.3)


Page 43 of 70

The energy policy is a cornerstone for implementing and improving an organisation’s EnMS and energy

performance within its scope and boundaries. The policy provides a statement of the high-level overview

of management’s intent that members of the organisation should apply to their work activities. The policy

also provides a framework for an organisation to set energy objectives and targets and associated energy

management action plans to further improve its energy performance. ISO 50001 requires an organisation

to at least state the following commitments in the energy policy:

● Continual improvement in energy performance;

● Availability of information and of necessary resources to achieve objectives and targets; and

● Compliance with relevant legislation and other requirements related to energy use, consumption

and efficiency.

In the context of implementing BAMA, in this step use cases are defined. In this step, based on the

expertise of employees and experts, areas are identified, where sensible improvements due to an

implementation of some or all of the BaMa modules can be expected. This already requires a significant

amount of knowledge about both the organisation and the capabilities of the BAMA tools.

Use Cases should be defined in a formal matter. In order to aid this process, two management tools where

developed throughout the course of the BaMa project. The first tool is a questionnaire. It helps experts to

identify and specify areas in which issues in a given company exist which might be solvable by the

application of BaMa tools.

Secondly, a template system specification document is provided which includes all relevant aspects

necessary to define a use case thoroughly. In this document, first the current state is described. Then

goals are formulated and finally the intended product with all relevant features are enumerated.

An especially important part of the system specification document are the metrics, with which the

achievement of the goals will be evaluated. Those metrics, among others, might both be used in an

optimization process (if necessary) and the other activities associated with the ISO 50001.

Once the definition of use cases is finished, the respective machines, which should be considered in order

to address them, need to be identified. Based on the already available data regarding those machines, the

number and nature of additional measurement locations can be defined. This often requires analysing

available data regarding energy consumption. Insights generated in this step will also help in the next

phase (energy planning).

As it is the case for most of the steps involved with the implementation of the BaMa tools, the system

specification is the most important document in this step. The goals determine not only what should be

measured, but also indirectly define how long additional measurements need to be conducted and how

detailed the generated data needs to be.

3.5.4 Conduct energy planning (Item 4.4)


It requires the organisation to identify and have access to the applicable legal and other requirements in

relation to its energy uses, consumption and efficiency to which it subscribes. (Item 4.4.2)


Page 44 of 70

It requires the organisation to develop record and maintain an energy review as well as document the

methodology and criteria used to develop the review. (Item 4.4.3)

It requires the organisation to establish an energy baseline(s) for the measurement of the energy

performance. (Item 4.4.4)

It requires the organisation to identify appropriate energy performance indicators to monitor and measure

its energy performance. (Item 4.4.5)

It requires the organisation to establish, implement and maintain documented energy objectives, targets

and action plans specified outcome or achievement defined to meet its energy policy related to improved

energy performance. (Item 4.4.6)

The element of legal and other requirements in ISO 50001 is intended to ensure that the organisation

complies with applicable legislation and other requirements related to energy use, consumption and

efficiency to which it subscribes. Legal requirements include those international, national, regional and

local governmental statutory requirements, which are applicable to the energy use of the organisation.

Other requirements refer to customers’ requirements, industry code of practices, government guidelines,

voluntary programs, public commitments of the organisation or its parent organisation, and requirements

of trade associations and others.

Regarding BaMa, based on the identified measurement locations, additional data acquisition has to be

done. This can for example include the measurement of energy demand throughout production processes,

production times, energy flows in energy systems or room temperatures. The measurements conducted

in order to identify and characterise the machines under consideration, can also be used within energy

reviews and therefore already serve multiple purposes.

In order to efficiently analyse measurements, relevant variables, parameters and metrics necessary to

monitor and control the processes within the production plant need to be defined. In the case of variables,

those often will coincide with the physical quantities measured. Parameters of those quantities often can

only be found after a preliminary analysis of the data, if respective expert knowledge is not readily

available. Regarding metrics, already defined ones can be adopted from the system specification. Vice

versa, not foreseeable metrics resulting from data analysis can be fed back into the systems specification.

As already mentioned, the measurement of energy flows will most likely be a part of the described

processes. Consequently, it is possible to evaluate the energy demand on machine level. This results in

an energy baseline for the respective parts of the organization, which in turn can then be used as part of

the energy planning described in ISO 50001.

Based on the relevant variables defined in the preliminary step, the respective simulation models can be

selected. If they already exist in a library, they can just be reused and the workload at setting up the virtual

representation of the production environment.

If the necessary models are not available, or existing models need to be adapted in order to fulfil the

requirements formulated in the system specification, the available model-base has to be extended. In order

to be able to do this, expert knowledge from the respective field is required. This can be provided by either

employees or external experts familiar with the respective domain.


Page 45 of 70

After the relevant energy flows, consumers (i.e. machines and equipment) have been identified and the

companies energy goals have been defined (à Item 4.4.2 and Item 4.4.6), the BaMa Tool can be modelled

and parametrized accordingly. Especially the BaMa optimization module is able to incorporate a

companies energy goals, relating to the available actuating variables. The energy goals from the energy

management thus have to be translated into a goal function (for more detail please refer to the

documentation on optimization).

An installed and working BaMa tool can support energy reviews (à Item 4.4.3) with measurable energy

efficiency improvement and a transparent model of a company’s energy consumption (à Item (à Item 4.4.4

and Item 4.4.5). The BaMa Tool and its underlying method are consistent with the demand for a clearly

defined and documented methodology as required by ISO 50001.

3.5.5 Implementation of energy management system (Item 4.5)


It requires the organisation to ensure all staff and persons related to significant energy uses are competent.

(Item 4.5.2)

It requires the organisation to address internal communication in relation to its energy performance and

EnMS. The organisation should also decide whether to communicate externally about its energy policy,

EnMS and energy performance. (Item 4.5.3)

It requires the organisation to establish, implement and maintain information to describe the core elements

of the EnMS and their interaction. (Item 4.5.4.1)

It requires to control all the EnMS documents. (Item 4.5.4.2)

It requires the organisation to identify and plan operations and maintenance activities, which are related

to its significant energy uses in order to ensure that they are carried out under specified conditions. (Item

4.5.5)

It requires the organisation to consider energy performance improvement opportunities and operational

control in the design of facilities, equipment, systems and processes that can have a significant impact on

its energy performance. (Item 4.5.6)

It requires the organisation to inform suppliers that procurement is partly evaluated on the basis of energy

performance when procuring services, products and equipment that have an impact on significant energy

use. (Item 4.5.7)

In order to implement the BaMa Toolchain successfully, a company has to provide a high level of data

quality and expert knowledge to model the relevant production system behaviour within the tool. This is

only possible with a competent team – a requirement consistent with Item 4.5.2. of ISO 50001. Since the

optimized planning provided by the BaMa Toolchain can only lead to real world benefits if the plans are

executed efficiently, it is of utmost importance that all staff are trained in the basics of industrial energy

efficiency generally and the system of energy goals of the company and the ways to achieve them in

particular. This necessitates training courses for a company’s workforce to be set up and conducted

regularly.


Page 46 of 70

Concerning the ISO requirements for communicating the energy performance and energy policy (Item

4.5.3) an implemented BaMa tool is able to provide transparency concerning the energy consumption of

a company and the improvements in energy efficiency in the domain of production planning and control.

The goals achievable through utilizing the BaMa Toolchain are a direct contribution to Item 4.5.6.

3.5.6 Checking (Item 4.6)


It requires the organisation to monitor measure and analyse the key characteristics of its operations that

determine energy performance at planned intervals. Equipment used in monitoring and measurement of

key characteristics should be calibrated to ensure data are accurate and repeatable. (Item 4.6.1)

It requires the organisation to evaluate compliance with legal requirements and other requirements. Which

it subscribes related to its energy use and consumption at planned intervals. (Item 4.6.2)

It requires the organisation to conduct internal audits regularly to ensure effective implementation of the

EnMS. (Item 4.6.3)

It requires the organisation to address nonconformities by making corrections, and by taking corrective

action and preventive action. (Item 4.6.4)

It requires the organisation to establish and maintain records to demonstrate conformity to the EnMS. (Item

4.6.5)

Concerning Checking, the BaMa Toolchain is related most of all to Item 4.6.1: Analysing the energy

performance of processes is greatly helped by the simulation based tool by providing detailed insight into

product and process specific energy consumption and how the energy consumption can be influenced

through optimized planning. To the remaining items in this section of ISO 50001 the BaMa method and

toolchain are 100 percent compatible although the requirements do not directly concern the BaMa method.

3.5.7 Management Review (Item 4.7)


It requires the top management to review the EnMS regularly to ensure its suitability, adequacy and

effectiveness. (Item 4.7)

Although the BaMa method and toolchain are compatible with ISO 50001 the management review, the

nature of a top management evaluation means only the top level decisions, guidelines, goal system

concerning the BaMa tool have to be considered. Within the management review, it is important to

establish, maintain and continuously improve the conditions for operating a planning tool such as the BaMa

toolchain effectively. The use of the BaMa toolchain provides detailed performance indication of a

company’s energy system and is thus contributing to efficiently and reliably providing the necessary

information for the review.


Page 47 of 70

4 Implementation level – “Provider”

4.1 Requirement definition

In order to be able to define the requirements towards the tool properly, a requirement specification

document should be created between the implementation and application partners. This document will

form the basis for the software implementation process as well as the final evaluation afterwards. Based

on the documents created in the course of the BaMa project, in the following section, the main sections of

such a document will be described. A template document will be available in the appendix.

4.1.1 Goals and Target Environment

In this section, based on a thorough analysis of the current situation in the project environment, the aims

of the project are described. In the description of the current situation, the application partner company

itself and its fields of operation can and should be described in this section. This context is necessary to

ensure that every partner has the same understanding of the project goals and consequently knows how

to address them.

Next, the initial situation should be described. Here, the relevant aspects of the boundary systems need

to be described as well as their shortcomings which motivated the project start in the first place. A thorough

description of those system is the basis for the formulation of aims and targets. The last part of this section

should be a description of the project instigation. In order to find solutions which are fitting to the application

partners needs, this can be extremely helpful.

Before the concrete aims are described, the desired situation after the project is finished should be

described in an general way. This approach helps stakeholders to envision the state of their company after

the project finished which not only is a good foundation for the formulation of well defined project goals but

also increases the motivation within the team.

Well defined goals help clarifying vague ideas and focus efforts. This leads to an efficient use of time and

resources. A good way to evaluate the quality of goals is the application of the SMART- Method. SMART

in this case is an acronym, combining attributes good goals should have.

Specific

A goal should be clear and specific, otherwise it won’t be possible to focus efforts. When drafting goals,

answering the five “W” questions (What, Why, Who, Where, Which) often proofs helpful.

Measurable

Measurable goals are important so the progress can be tracked. It also is the prerequisite for the evaluation

of results and therefore the determination of success.

Achievable


Page 48 of 70

Goals need to be realistic. This does not mean that they should not be ambitious in the sense that the

stretch the abilities of the project team, but in order to keep the team motivated, goals need to be

achievable. This already needs a good common understanding of the problem domain and competence

within the project team as this is the prerequisite for assessing the workload associated with endeavours

correctly.

Relevant

This step is meant to ensure, that the project and its results are worth the effort. The project should align

with other relevant goals and everyone involved in the project is responsible for the expected results and

their implications.

Time bound

Every project needs a target date. Deadlines help to focus team efforts and therefore improve the overall

product quality.

Another way to ensure high goal quality, the formulation of non- goals and optional goals as well as the

distinction between system- goals and process- goals might be helpful in order to achieve more clarity.

Finally, in order to check the validity of the formulated goals, the expected result and its usefulness should

be described. The most important aspects of the result, being the areas of application and the targeted

user groups should also be described as precise as possible. A good understanding of the users is a

prerequisite for a successful implementation.

4.1.2 Product Use

The most important aspects of the result, being the areas of application and the targeted user groups

should also be described as precise as possible. Within the area of application section, the processes

within which the product should be applied are laid out. Example processes in the BaMa context might be

documentation, Energy reviews or production planning. By describing the intended user group, important

context in regards to required background knowledge or tool literacy is given. A good understanding of the

users is a prerequisite for a successful implementation.

4.1.3 Product Overview

In the product overview, the users are described in a more technical manner in the description of the user

structure. This defines the user groups and their roles when interacting with the product. Based on the

definition of the user-groups, a comprehensive list of features including a short description as well as the

context (users, prerequisites, results…) should be written. The interaction between those features and the

different user groups can then optionally be visualised using UML- Use Case diagrams.


Page 49 of 70

Figure 25 User groups

The interaction between those features and the different user groups can then optionally be visualised

using UML- Use Case diagrams.


Page 50 of 70

Figure 26 Use case diagram

4.1.4 Product description

In the project overview, the concrete product, which will be developed in the course of the project, is

described in a more detailed manner. Based on the nature of the project, this description could for example

include figures regarding the software architecture or mock-up user interfaces.


Page 51 of 70

Figure 27 Mock up GUI

4.2 Implementation of the optimization framework

4.2.1 Implementation of the optimization framework - Vienna University of Technology

The optimization framework consists of the following parts:

● Hybrid simulation module (from AutomationX and Vienna University of Technology)

● Scenario: The production scenario comprises an initial production plan, a demand plan and a

chosen season (weather influence)

● Input file: this parameterized file includes the concrete information for the given scenario and is

called by the simulation module, the file is already filled with a valid initial solution from the

Production plan-Generator module

● Results: The results from one simulation run serving as feedback from the simulation module

● Target function: Evaluates the results from each simulation run and represented by one → “fitness

value”, giving a final information about the quality of the actual solution

● Production plan-Generator: This module generates a valid production plan out of the chromosomal

structure of the storage (and oven) vectors → ready to fill the input-file and finally ready to call the

simulation again, see the according figure above


Page 52 of 70

Figure 28 structure SEPPI optimization

Figure 29 Production plan generator

Originally, the optimization deals with a “Permutation Flow Shop Scheduling” Problem of an industrial

bakery (MPreis, Völs). In order to handle this scheduling problem, the “Genetic Algorithm” (GA) from the

Matlab Global Optimization Toolbox was found to suit best to the given problem and implemented in a

customized variant. This metaheuristic calls the hybrid simulation module with adapted production plans,

the simulation module evaluates those plans, giving results that are evaluated by the target function. The

GA works by selecting a mixture of more and less fitter individuals (represented by their chromosomal

structure) to generate a new set of different individuals.

Tuning and enhancement of the standard algorithm

Tuning achieved by:

● parameterization (comprehensive tests of the Genetic Algorithm options)

● guided search mechanisms by customization of the specific Genetic Algorithm operators

● 2-phase-GA-approach:

o phase 1: Optimized production sequence

o phase 2: Optimized energetic consumption by contracting, shifting and swapping oven

uptimes

Enhancements:

● Integer Optimization


Page 53 of 70

● Memory Function

● Hybridization with other (meta) heuristics → Matlab: Pattern Search

Scenarios

We have derived three scenarios (1 day, 1 week, 1 month) with each of them being represented by two

demand plans. The two demand plans have variations concerning their demand on product type and

corresponding quantity. This documentation guideline focuses on the description of the more

comprehensive second prototype of the production line 3 of MPreis.

Target function

The target function covers part-goals concerning delivery tardiness and storage (f1 and f2) as well as

energy costs and penalties for CO2 and demand plan deviations. Finally, also the oven-uptimes and lot-

changeover-costs have been added into the target function. The structure of the implemented target

function is given on the figure above:

The delivery tardiness is evaluated on product type level, while storage costs are evaluated on production

lot level. The next figure shows the inclination of each part-function and the dependencies between those

two part functions.

Figure 30 Target function of the second prototype


Page 54 of 70

Figure 31 Part goals representing delivery tardiness and storage costs

In addition to the structure of the implemented formula, it has to be mentioned:

● Different delivery tardiness penalties for fresh (100€/palette) and frozen (20€/palette) products

● Doubled storage costs for frozen products to simulate the higher acquisition costs of the frozen

storage and the higher storage costs itself

● oven-uptime costs: 100€/hour

● modified (virtual) lot changeover-costs, see figure below:

Figure 32 Lot changeover-cost-matrix

Constraints

Constraints represent very important limiters in complex optimization problems and have to be set up

according to the production planning problem to reduce the given search space and to ensure valid

solutions. We have made the setup within each optimization phase. In optimization phase 1, we have

realized one simple constraint to ensure positive vectors concerning production start. The other constraints

were more complex and considered in optimization phase 2 (see figure below). The first constraint

constants that the end of the previous oven production lot has to be finished before the oven aggregate

can be started for the next oven lot. The second constraint claims that the oven has to run for each

timeframe (as minimum) the time-length of the smallest lot within the given scenario. The third constraint


Page 55 of 70

ensures that the oven has to run at least the sum of all oven occupation-times of all lots of the given

scenario. These constraints ensure valid solutions in accordance with the oven aggregate.

Figure 33 Constraints

Figure 34 Legend

Algorithm tuning

In order to achieve good simulation results the algorithm has to customize. We decided the customization

to result in a guided search represented by the GA operators. We customized the operators to perform

● guided swapping of lots: specific lots are swapped and re-scheduled → enable guided sequencing

● guided mutation of lots: specific lots are mutated in a guided way to optimize the start-of-production

(SOP) times

● guided oven contractions: specific modified oven turn-on and turn-off times (delayed oven-starting

times and premature oven-stop times); in addition oven lots are shifted within a certain time-period

Simulation results

The simulation results deliver the optimization potential and depend on the tuning and implementation of

the corresponding optimization algorithm. The next figure shows a typical Matlab plot, consisting of nine

sub-figures showing the number of variables within the specific production scenario next to other scores.


Page 56 of 70

Figure 35 Typical Matlab plot within optimization

The following results are delivered by simulation studies of different scenarios. The next figure shows the

global goal optimization scores (~ 50%) achieved within a 1-day-scenario of one production line.

Figure 36 global goal line 3 1day

The next figure shows the season influence within 1 production day (same production volumes, stocks

and demand plans - but another “day” in a different season). Season 1 represents week number 03

(winter), season 2 represents week number 34 (summer) and season 3 constitutes week 44 (autumn).


Page 57 of 70

Figure 37 energy optimization season influence 1day

The next figure shows the part-goals development within the global goal optimization. An important

annotation is the guided energy optimization in phase 2.


Page 58 of 70

Figure 38 part goals development within optimization

The next figures show the additional benefits by comparing different GA population sizes and the optimal

integer step value (within an integer optimization). The final optimization results deliver significant saving

potentials from 20 - 50%, depending on the production scenario.

Figure 39 population size tests (7-day-scenario)


Page 59 of 70

Figure 40 Integer optimisation comparison (line 3, 7-day-scenario)

Within the optimization there were different product types with different demands, we differentiated

between products with daily demands, weekly demands and demands twice a week. The following

example gives an overview about the capabilities of the implemented sequencing module in order to

achieve a solution with minimized product-type specific changeover costs, the changed and optimized

sequence parts by the optimization algorithm are marked in red colour:

Initial sequence:

7 – 6 – 5 – 6 – 5 – 1 – 3 – 5 – 6 – 5 – 6 – 3 – 1 – 5 – 6 – 5 – 3 – 1

Optimized sequence –run 1:

5 – 6 – 6 – 1 – 1 – 1 – 3 – 6 – 5 – 5 – 5 – 3 – 7 – 5 – 5 – 6 – 6 – 3

Optimized sequence –run 2:

6 – 1 – 1 – 1 – 5 – 5 – 3 – 5 – 5 – 6 – 6 – 6 – 6 – 7 – 5 – 5 – 3 – 3


Page 60 of 70

Technical building equipment optimization

As a last step a dedicated (third) optimization phase was performed after the two-phase approach in order

to achieve an optimized utilisation of the technical building equipment. Within this phase, the chillers

received a specific thermal optimization, whereby the priority of the chiller usage can be adjusted within

the input file for the simulation. The target function for this optimization represented the energy

consumption, for optimization the standard Genetic Algorithm with only swap-mutation (of the chiller

aggregates) was activated for the solution development process (-> optimized chiller usage during a

planning horizon). The optimization results identified another 5% energetic potential, see also the following

figure.

Figure 41 3-phase-optimization

4.2.2 Implementation of the optimization framework - AutomationX

The problem is to minimize some weighted costs of production of a known number (n) of products for the

bakery MPREIS. This is a NP hard problem.

We represent the sequence (order) of batches {B0 … Bn-1} as a permutation π of integers from 0 to n-1.

Between each two consecutive batches there is a gap gi >= 0 (i >=0, i<N). Batches are determined by

their start-time on storage cube.


Page 61 of 70

Figure 42 Optimization scenario

When all gaps are equal to zero, we have a tight schedule.

All cube instances (up to oven-type) of the axdevs modelling of the Line3 of MPREIS have a known static

behaviour. Each cube can calculate/simulate for each batch the incoming and outgoing time at cubes

interfaces. Cubes of oven-type need a settling time (ovstm) for product-baking temperature.

This is dependent on product type, the current temperature of the cube and external factors like ambient

temperature.

If the gap between two batches is large, the oven cube will set to standby mode to save energy. Therefore,

we define a beginning of the gap (g) and an oven pause parameter (ovp) with the constraint: ovp + SetTm

<= g:

Figure 43 Optimization step 1

Axdevs-simulation time for 32 batches, 7-day scenario on Line3 is approximately 4 seconds, if parallel

simulations (multi thread) mode is active. This lasts a long time for a GA-solver therefore we are confined

to work with small population size to moderate iteration count. Invalid schedules are another big problem,

it may happen, because of congestion, that not all of the products will be produced. In this case the GA

attempts to repair the existing schedule, or creates a new schedule. Both actions are time-consuming.

To solve this problem every single batch will be simulated in advance of the production, to receive the

required information like, number of entities produced for each batch, arrival time of the first product, the

time when the last product of this batch left the cube, and many more. For each oven cube the temperature

profile will be calculated. Using cheap interpolation techniques in these profiles we can estimate the

temperature of oven after the break (ovp) and the temperature after the settling times (ovstm).


Page 62 of 70

With this information, valid schedules can be created for the GA.

Objective function

F(x) = ω1 *T1+ω2 * T2+ω3*T3 where:

x: free variables -> {π, g1,..gn}

T1: cost for storage / delayed delivery (summed up for all jobs)

T2: cost for production-line usage (proportional to total flow time)

T3: cost for energy (including CO2 )

ωi>= 0 are weighting factors.

In the case of ω3 = 0 the objective function can be calculated without simulation. In this case larger

population sizes and higher iteration counts can be used, because the energy term (T3) is rather smooth

in this case and can be used as a surrogate objective function.

● Chromosome: free variables -> permutation π, gaps gi and ovpi.

● Selection: proportional roulette wheel.

● Crossover: order based crossover, select randomly two parent chromosomes and several positions

from range [0, N-1] and impose the order of batches at positions in one parent to the other parent.

Genetic operators.

shift: pick a random position i from [0..N-1] and a random ∆t apply gi+= ∆t -> start times for Bj (j >i) will be

shifted for ∆t.

modify gap : pick a random position i from [0..N-1] and a random ∆t apply gi += ∆t -> start times for Bj (j

>i) will be shifted only if ∆t > gi+1



Page 63 of 70


Transposition (T): pick two random positions i,j (i<j) from [0..N-1] and swap batches at positions i,j.


In case when j = i+1 the transposition is a simple one (ST).

If the calculation of objective function is low then it is better to use ST, it sets cardinality, is smaller (n-1)

and induces a higher search space diameter than T.

4.3 System requirements and Hardware

4.3.1 Measurement Structure

In order to provide a sound basis for energy efficiency measures in general and cube identification in

particular, transparent energy flows and thus energetic measurements are recommended. If the measured

data are centrally recorded, visualized, processed and stored over a longer period, this is referred to as


Page 64 of 70

energy monitoring. The classic monitoring approach is to locate it at company level, in order to determine

the consumption of the entire plant and of certain distributors (halls, areas and / or cost centres).The

roughly quantified consumptions can be recorded in tabular form and then be ordered and prioritized

according to their consumption quantities (ABC analysis / pare diagram).

Figure 47 Measurements equipment

In the actual design, the first step is the definition of the number and position of the measuring points. The

relevant outputs from the transformer / main distributor are the first sensible measuring points in the case

of electrical energy, whereby the connection values (easy to recognize at the used fuses) help in the

selection. From here, the degree of detail can be increased in the top-down approach, i.e. in sub-

distributors or control cabinets of specific technical consumers (see Figure 48).

Figure 48 Energy balances

The energetic behaviour of uninfluenced consumers or business areas can also be determined by means

of an intelligent selection of the measurement points by way of differential calculations, which are then so-

called virtual measuring points.

Since in the given case there are already assumptions regarding the main users, it is useful to check their

actual energetic behaviour by means of short-term measurements. If the measurement period is selected

long enough, these temporary analyses can often lead to interesting findings regarding the usage profile

of the respective consumer, and possibly pre-emptive assumptions.


Page 65 of 70

In the case of a measuring device with a data logger, it is recommended to measure the measurement

time for at least one week or to measure it at several different points in time, since the operating behaviour

of the consumer can also be analysed during the working time. Thus, for example, consumers with low to

medium power consumption can show that they have higher consumption rates due to a high running time

(or even continuous operation) than consumers with high power consumption and relatively short operating

times.

In order to identify the respective energetic hotspots for the individual energy carriers, the energy portfolio

analysis is suitable. It categorizes energy consumers into critical and (for the time) uncritical consumers.

Figure 9 gives an overview of the so-called energy portfolio and corresponding recommendations for

different types of consumption.

Figure 49 Machine categorization

Based on connection values and durations, a distinction is made between four categories of consumers.

Critical consumers are immediately put into focus for more in-depth microanalysis.

Electrical Power Monitoring

The market provides a large number of devices for measuring the measurement of energy data. The

selection of the devices is in principle the responsibility of the operator and must be adapted to the

individual requirements. It can roughly be distinguished between devices for mobile use and permanently

installed measuring devices.


Page 66 of 70

The multimeter data logger is particularly suitable for mobile, temporary measurements in the operational

environment. During the measurement in the three-phase alternating circuit, separable measuring clamps

(current transformers) or Rogowski coils are used in connection with voltage clamps. From the measured

current amplitudes and voltages, certain quantities are calculated by the processor after device-internal

signal conversion. A software or manufacturer-specific software is usually used for this purpose. The

measured values can be read directly on the device, stored in the data logger and / or sent to a PC.

The following exemplary solutions (multi-functional devices, see Figure 11, or module-based systems) for

three-phase power measurements have proven themselves industrially (communication interface and

gross list price are given):

● Siemens SENTRON PAC 3100, MODBUS RTU, EUR 270,-

● Siemens SENTRON PAC 3200, MODBUS TCP, EUR 450,-

● Weidmüller POWER MONITOR 51A, MODBUS RTU, EUR 540,-

● Weidmüller UR20 Energiemessklemme mit Buskoppler, MODBUS TCP, EUR 460,- (ohne 24 V

Stromversorgung)

● ALGODUE UPM307, MODBUS RTU, EUR 400,-

● Janitza UMG 96S, MODBUS RTU, EUR 250,-

● GMC SINEAX A210, S0, EUR 310,-

● B&R X20 Energiemessmodul und Buskoppler, MODBUS TCP, EUR 250,- (ohne 24 V

Stromversorgung)

● Phoenix Contact Inline-Funktionsklemme und Buskoppler, MODBUS TCP, EUR 580,- (ohne 24 V

Stromversorgung)

Figure 50 Electrical power measurement

In addition, test leads must be provided for the voltage reduction and current transformers. For the latter,

for example, ELEQ systems are recommended for primary currents of 60-1000 A. Depending on the

primary current, the costs per converter are max. around EUR 100, -.

The cost per measuring point can be reduced if, instead of a three-phase measurement, a single-phase

measurement is implemented or even a current measurement is performed. For this purpose, however, in

the first case, in the first case, in order to avoid major errors, it is expedient to verify the symmetry of the

load by way of a three-phase measurement or, in the second case, to determine the phase angle or power

factor. Furthermore, in the first case, it is generally assumed that the loads to be measured are located in

the same distribution cabinet, since otherwise complex cabling to the measuring devices is necessary.


Page 67 of 70

Another strategy would be the measurement using classic built-in energy meter solutions, which often

have a lower accuracy and require an interruption of the circuit, but do not require a converter and are

more cost-effective, thus greatly reducing overall costs.

Gas flow measurement

Depending on the pipe dimensions and accuracy requirements, various systems are available for

monitoring gas consumption. An exemplary system is the thermal flow sensor HAUTZSCH TA10M-265G

(see Fig. 12), whose price per measurement point amounts to around 2200. The system has an M-Bus

interface. Alternative, similar systems are offered by manufacturers such as Molline or Voeglin.


Page 68 of 70

LITERATUR

[1] Eurostat, “Energy balances.” [Online]. Available:

http://ec.europa.eu/eurostat/web/energy/data/energy-balances. [Accessed: 22-Jul-2015].

[2] E. Bonneville and A. Rialhe, “Good practice for energy efficiency in industry,” 2006.

[Online]. Available: https://leonardo-energy.org/sites/leonardo-

energy/files/root/Documents/2009/DSM-industry.pdf. [Accessed: 15-Nov-2012].

[3] M. Paulus and F. Borggrefe, “The potential of demand-side management in energy-

intensive industries for electricity markets in Germany,” Appl. Energy, vol. 88, no. 2, pp.

432–441, Feb. 2011.

[4] G. Chryssolouris, Manufacturing Systems: Theory and Practice, 1st ed. New York: Springer

Science & Business Media, 1992.

[5] K. Bunse, M. Vodicka, P. Schönsleben, M. Brülhart, and F. O. Ernst, “Integrating energy

efficiency performance in production management – gap analysis between industrial needs

and scientific literature,” J. Clean. Prod., vol. 19, no. 6–7, pp. 667–679, Apr. 2011.

[6] K. Salonitis and P. Ball, “Energy Efficient Manufacturing from Machine Tools to

Manufacturing Systems,” Procedia CIRP, vol. 7, pp. 634–639, 2013.

[7] V. Introna, V. Cesarotti, M. Benedetti, S. Biagiotti, and R. Rotunno, “Energy Management

Maturity Model: an organizational tool to foster the continuous reduction of energy

consumption in companies,” J. Clean. Prod., vol. 83, pp. 108–117, Nov. 2014.

[8] F. Alhourani and U. Saxena, “Factors affecting the implementation rates of energy and

productivity recommendations in small and medium sized companies,” J. Manuf. Syst., vol.

28, no. 1, pp. 41–45, Jan. 2009.

[9] J. R. Duflou, J. W. Sutherland, D. Dornfeld, C. Herrmann, J. Jeswiet, S. Kara, M. Hauschild,

and K. Kellens, “Towards energy and resource efficient manufacturing: A processes and

systems approach,” CIRP Ann. - Manuf. Technol., vol. 61, no. 2, pp. 587–609, Jan. 2012.

[10] S. Rahimifard, Y. Seow, and T. Childs, “Minimising Embodied Product Energy to support

energy efficient manufacturing,” CIRP Ann. - Manuf. Technol., vol. 59, no. 1, pp. 25–28,

2010.

[11] S. Thiede, G. Bogdanski, and C. Herrmann, “A Systematic Method for Increasing the

Energy and Resource Efficiency in Manufacturing Companies,” Procedia CIRP, vol. 2, pp.

28–33, 2012.

[12] C. Herrmann, S. Thiede, S. Kara, and J. Hesselbach, “Energy oriented simulation of

manufacturing systems – Concept and application,” CIRP Ann. - Manuf. Technol., vol. 60,

no. 1, pp. 45–48, 2011.

[13] S. Kara and W. Li, “Unit process energy consumption models for material removal

processes,” CIRP Ann. - Manuf. Technol., vol. 60, no. 1, pp. 37–40, 2011.

[14] N. Weinert, S. Chiotellis, and G. Seliger, “Methodology for planning and operating energy-

efficient production systems,” CIRP Ann. - Manuf. Technol., vol. 60, no. 1, pp. 41–44, 2011.

[15] I. Kovacic, K. Orehounig, A. Mahdavi, F. Bleicher, A.-A. Dimitrou, and L. Waltenbereger,

“Energy Efficient Production – Interdisciplinary, Systemic Approach through Integrated

Simulation,” Stroj. J. Theory Appl. Mech. Eng., vol. 55, no. 1, pp. 17–34, Feb. 2013.


Page 69 of 70

[16] F. Bleicher, F. Duer, I. Leobner, I. Kovacic, B. Heinzl, and W. Kastner, “Co-simulation

environment for optimizing energy efficiency in production systems,” CIRP Ann. - Manuf.

Technol., vol. 63, no. 1, pp. 441–444, 2014.

[17] J. Hesselbach, C. Herrmann, R. Detzer, L. Martin, S. Thiede, and B. Lüdemann, “Energy

Efficiency through optimized coordination of production and technical building services,”

Jan. 2008.

[18] P. D. Ball, M. Despeisse, S. Evan, R. M. Greenough, S. B. Hope, R. Kerrigan, A. Levers, P.

Lunt, M. R. Oates, L. Shao, T. Waltniel, C. Wheatley, and A. J. Wright, “Modelling buildings,

facilities and manufacturing operations to reduce energy consumption,” in Proceedings of

the POMS 23rd Annual Conference, 2001.

[19] S. Hu, F. Liu, Y. He, and T. Hu, “An on-line approach for energy efficiency monitoring of

machine tools,” J. Clean. Prod., vol. 27, pp. 133–140, May 2012.

[20] J. L. Michaloski, G. Shao, J. Arinez, K. Lyons, S. Leong, and F. Riddick, “Analysis of

sustainable manufacturing using simulation for integration of production and building

service,” pp. 93–101, Apr. 2011.

[21] Y. Liu, “Dynamic study on the influencing factors of industrial firm’s carbon footprint,” J.

Clean. Prod., vol. 103, pp. 411–422, Sep. 2015.

[22] L. Čuček, J. J. Klemeš, and Z. Kravanja, “A Review of Footprint analysis tools for

monitoring impacts on sustainability,” J. Clean. Prod., vol. 34, pp. 9–20, Oct. 2012.

[23] J. P. Padgett, A. C. Steinemann, J. H. Clarke, and M. P. Vandenbergh, “A comparison of

carbon calculators,” Environ. Impact Assess. Rev., vol. 28, no. 2–3, pp. 106–115, Feb.

2008.

[24] M. Gaussin, G. Hu, S. Abolghasem, S. Basu, M. R. Shankar, and B. Bidanda, “Assessing

the environmental footprint of manufactured products: A survey of current literature,” Int. J.

Prod. Econ., vol. 146, no. 2, pp. 515–523, Dec. 2013.

[25] T. Wiedmann and J. Barrett, “A Review of the Ecological Footprint Indicator—Perceptions

and Methods,” Sustainability, vol. 2, no. 6, pp. 1645–1693, Jun. 2010.

[26] E. Lindskog, L. Lundh, J. Berglund, Y. Tina Lee, A. Skoogh, B. Johansson, and B.

Johansson, “A method for determining the environmental footprint of industrial products

using simulation,” in Proceedings of the 2011 Winter Simulation Conference (WSC), 2011,

pp. 2131–2142.

[27] G. May, I. Barletta, B. Stahl, and M. Taisch, “Energy management in production: A novel

method to develop key performance indicators for improving energy efficiency,” Appl.

Energy, vol. 149, pp. 46–61, Jul. 2015.

[28] Rager, M., 2008. Energieorientierte Produktionsplanung: Analyse, Konzeption und

Umsetzung. Springer-Verlag.

[29] Eiben, A.E., Smith, J.E., 2015. Introduction to evolutionary computing, 2nd edn. Springer,

Berlin.

[30] EPEX SPOT. Marktdaten Day-Ahead-Auktion: Börse für Stromspotmärkte.

https://www.epexspot.com/de/marktdaten/dayaheadauktion/auction-table/. [31] Österreichisches Institut für Bautechnik. OIB-RICHTLINIE 6 - Energieeinsparung und

Wärmeschutz(OIB-330.6-009/15).

https://www.oib.or.at/sites/default/files/richtlinie_6_26.03.15.pdf.

https://www.epexspot.com/de/marktdaten/dayaheadauktion/auction-table/

https://www.oib.or.at/sites/default/files/richtlinie_6_26.03.15.pdf


Page 70 of 70

[32] European Energy Exchange (EEX). European Emission Allowances | Global Environmental Exchange: Preis in EUR/tCO2. https://www.eex.com/de/marktdaten/umweltprodukte/spotmarkt/european-emission-allowances#

Contact data

Project lead: Prof. Dr. Friedrich Bleicher

Institute: Institute for production engineering and laser technology, TU Wien

Getreidemarkt 9, 1060, Vienna, +43 1 58801 31101, [email protected], http://www.ift.at,

http://bama.ift.tuwien.ac.at

https://www.eex.com/de/marktdaten/umweltprodukte/spotmarkt/european-emission-allowances