Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 6

Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 99 (2021) 9–14
www.elsevier.com/locate/procedia

14th CIRP Conference on Intelligent Computation in Manufacturing Engineering, CIRP ICME ˈ20

Generative design in factory layout planning


Marian Süßea*, Matthias Putza
a
Fraunhofer Institute for Machine Tools and Forming Technology IWU, Reichenhainer Straße 88, 09126 Chemnitz, Germany

* Corresponding author. Tel.: +49 371 5397-1517; fax: +49 371 5397-6-1517. E-mail address: marian.suesse@iwu.fraunhofer.de

Abstract

Planning and optimization of facility layouts have been investigated for decades and manifold approaches are applied for structuring and design
of production layouts. However, results heavily depend on the experience and creativity of involved planning experts. Currently, complexity of
planning processes constantly increases, e.g. due to further requirements of energy and media supply. Generative Design, hitherto mainly
applied in component development, provides opportunities to cope with a much larger solution space and develop creative layout concepts.
Thus, based on a structured overview on established planning methods, a concept and first results of factory layout planning with Generative
Design are described.
© 2021 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 14th CIRP Conference on Intelligent Computation in Manufacturing Engineering,
15-17 July 2020.
Keywords: Generative design; Computational design; Evolutionary design; Factory planning; Layout planning; Literature search

1. Introduction solution spaces. Just as the environmental and resilience


scientist Marten Scheffer [2] reflects, “[…] the best science
Factory planning may be defined as a structured process tends to come from a balanced mix of rationality and
that serves for developing a target-oriented design of adventurous association.” This enforces the assumption that
production systems and their according infrastructure to GD approaches may provide promising opportunities in
efficiently generate developed goods and services. Thus, combination with the well-structured factory planning
multidimensional decisions need to be made in procedures. Thus, this paper will investigate its previous and
interdisciplinary teams. The applied techniques may range concurrent application for factory planning tasks with focus
from intuitive and experience-based design solutions to on layout planning.
sophisticated methods and optimization approaches. In Therefore, main phases and decisions to be made in factory
addition to that, the design of the factory’s structure or layout layout design will be described in the subsequent section. This
plays a predominant role as it defines major boundaries for also comprises the status of available software for layout
further planning stages. As shown by Rist [1], the planning of planning in manufacturing. Based on this, a short
a factory’s structure comprises the most interfaces towards characterization of GD is given. In the third section, results of
further distinct planning topics. a structured literature research, its limitations and results of
Generative Design (GD) gained attention as an approach major relevance will be presented. Afterwards, challenges for
that makes use of evolutionary principles for development the application of GD in factory layout planning are derived
processes in manufacturing and engineering. At a first glance and a first framework for planning and decision-making with
these methodologies are mainly known as part of heuristics GD techniques is shown. The final outlook explores further
for solving optimization problems. However, due to the research directions.
increasing computer performance and improvements in
software design and programming environments the bio-
inspired principles find further application for the design
and analysis of large

2212-8271 © 2021 The Authors. Published by Elsevier B.V.


This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 14th CIRP Conference on Intelligent Computation in Manufacturing Engineering,
15-17 July 2020.
10.1016/j.procir.2021.03.002
10 Marian Süße et al. / Procedia CIRP 99 (2021) 9–14

2. Scope of factory layout planning & Generative Design design software still uses these simple and outdated methods
in the applications.” Based on their findings several reasons
2.1. Processes and decisions in layout planning hinder the progress. On one hand, it may be the lack of
knowledge in theory and application of advanced scientific
The factory planning process has been formalized and optimization approaches among both, potential software users
described in manifold variations and comprises several but also software developers. On the other hand, low
hierarchy levels of a production site. The processes and acceptance among the involved planners occurs due to
decisions described in this paper refer to the arrangement of increased automation that seemingly shifts power of decision
elements (i.e. areas, machines etc.) in one building. to IT departments. Besides that, a high number of target
A standardized procedure in accordance with the criteria and boundaries, especially in the latter planning
established official scale of fees for services by architects and phases, appear to be a further cause for the gap between
engineers (HOAI) [3] is described in the norm VDI 5200 by theory and implementation. This accounts even more in
the German Society of Engineers [4]. With reference to this brownfield scenarios when further restrictions flank the
basic description, several authors provide systematization of planning process.
the overall procedure and its separate steps with varying foci
[5– 8]. A consolidating overview of the phases is given by 2.2. Introduction to Generative Design
Clauß [9], who separates them into (1) preparation, (2)
structure planning, (3) detailed planning, (4) execution Progress in hardware performance but also an increasing
planning and (5) execution. Decisions regarding the structure number of CAD software solutions during the last years
and production layout may generally be situated in the second reduce access hurdles to programming and optimization for
and third phase. many disciplines like engineering or architecture. In a similar
Overall target of these planning stages is a good, or even manner GD has already been characterized by Nagy et al.
optimized, arrangement of the factory’s facilities. Common [15], who stated that “Generative Design allows designers to
principle is the elaboration and assessment of different tap into the power of computation to explore large design
solution alternatives in order to identify a preferred variant spaces and derive design solutions which are both novel and
which suffices quantitative and qualitative criteria. Hence, high- performing relative to a chosen set of goals.” However,
there are manifold definitions of one ore multidimensional nature- inspired (e.g. evolutionary or swarm-based) principles
targets (e.g. minimized transportation costs, maximized that are applied in the engineering domain are by no means
flexibility). Often, they are coped with intuitive decision novel approaches but based on the indicated progresses they
making based on the planners’ experience, individual may support planning by facilitating optimization and
predilection or spreadsheet- based multicriteria decision variation.
making with scoring models or the Analytic Hierarchy Stouffs and Rafiq [16] substantiate this by stating that the
Process (AHP). With regard to this, a large toolset of optimization techniques in GD are mainly applied to provide
modelling approaches and solution techniques has been insight into a large set of solutions (i.e. the solutions space)
developed in mathematical optimization and Operations which allows the consideration of a multiple amount of
Research where these investigations are subsumed in the alternatives compared to manual decision methods. Due to
facility layout planning problem (FLP) and its variations. A that, main aspect of the investigation is not a global optimum
structured literature overview which comprises an analysis of or the optimization technique itself but their application to
layout problems, their mathematical representation and support planning and design by some type of structured
formulation as well as resolution approaches is provided by creativity. The genesis of design methodologies in different
Drira et al. [10]. An extensive investigation with even more engineering fields, and therefore the dissociation of GD has
focus on its optimization techniques has been conducted by been summarized by Wypch and Wach [17] in a presentation
Anjos and Vieira [11]. at the Autodesk University 2019 in Las Vegas and is depicted
Some heuristics like the Hungarian method [12] for solving in the following Figure 1.
the FLP as an assignment problem were made accessible for Due to its widespread use, a structured literature analysis of
factory planners and approaches for layout generation on a GD use cases, the scientific domains and any relation to the
rough level. Its derivations like the solving method of described factory layout planning processes is conducted in
Schmigalla are partly established for manual layout the subsequent chapter.
optimization in early planning phases [7].
The available software solutions for factory modeling and
visualization are steadily increasing but the transition from
theoretical progress and prototype implementation into
software is missing. In an exemplary case of Banduka et al.
[13] layout modelling is executed by manual application of
the triangle-based arrangement according to Schmigalla in the
specified software visTABLE. Different layout alternatives
are only generated based on the planner’s creativity and
experience. In a specific investigation of layout planning
solutions, combining market research and semi-structured
interviews, Lohmer et al. [14] emphasize that “[…] current
Figure 1: Genesis of design concepts (following Wypch and Wach [17])
3. Structured literature search
Naturally, this growth rate on both databases is not only
3.1. Overview on search strategy and results determined by increased relevance of the topic itself, but even
more by the spread of the world wide web and digital access
Since current availability of online resources and databases to research.
leads to an enormous and steadily increasing amount of Compared to that, first papers for the search terms related
scientific papers in many different fields, it is much more to GD were constantly published from the early-1970s on
necessary to provide a comprehensible way for the selection lower rate compared to layout planning. A steady rise in
of relevant previous work. Therefore, the two described “SCOPUS” starts in 1996 which continuously exceeds results
research areas in section 2 were primarily subdivided into in layout planning from 1999 onwards. The results in “Web of
similar and accompanying terms which has led to the Science” show a parallel development with a strong leap in
selection in Table 1. Conducted in March 2020, the literature 2013 (see Figure 2). The strong incline of GD in both
search was limited by 2019 because (less) findings of the databases may be caused by the relevance of its principles in
ongoing year would distort quantitative comparisons. “Web multiple areas. As an example, the results for GD in “Web of
of Science” and “SCOPUS” were selected as two different Science” are spread out over 146 categories from which
interdisciplinary databases. The search terms per column were “Biochemistry Molecular Biology” (15,8%), “Computer
connected by the OR-operator and each topic was initially Science Interdisciplinary Applications” (12,3%), “Chemistry
investigated independently in order to observe the Multidisciplinary” (11,9%), “Architecture (7,6%) and
appearance of research “Material Science Multidisciplinary” (6,9%) form slightly
related to times. more than the half of the results. “Engineering
Multidisciplinary” is listed with 6,2% on 7th place. On the
Table 1: List of search terms and accumulated results
contrary, the results for factory layout planning comprise 118
Topic Generative Design Layout Planning
science categories of which
Search Terms generative design “Engineering Industrial” (13,4%), “Operations Research
layout planning
computational design factory planning Management Science” (19,3%) and “Engineering
evolutionary design Manufacturing” (14,8%) already amount to more than the half
space planning
of the total quantity.
factory layout planning
A further search was conducted by combining both lists
facility layout planning with the AND-operator, as shown in the following overall
facility layout problem search strategy: (“generative design” OR “evolutionary
design” OR
SCOPUS 5,645 results 3,589 results “computational design”) AND (“factory planning” OR
Web of Science 4,875 results 1,665 results
“layout planning” OR "space planning" OR “factory layout
planning” OR “facility layout planning” OR “facility layout
problem”). The combined search produces 15 results in
Among both databases the first identified research paper in “SCOPUS” and 8 results in “Web of Science”. Among the
Layout Planning dates back to 1955 and was published by latter, only one paper was not contained in the “SCOPUS”-
Jones [18]. Interestingly, it bears the promising question: “Is results. Moreover, one of these results only reflects general
research necessary for equipment and layout planning?” and conference proceedings without a specific paper. This leads to
refers to the planning of hospitals and medical departments. an overall literature fund of 15 articles.
Although publications for factory layout planning These results were concentrated on publications that are
continuously arise from the mid-1960s, the number constantly linked to all scientific areas tagged with “engineering”. Thus,
fluctuates around 30 results per year until the end of the 20 th the work of Jo and Gero [19], Rosenman and Saunders [20,
century. A continuous growth in “SCOPUS”-results can be 21], Fernando [22] and Nagy et al. [23, 24] were selected for a
seen from 2002 onward and in “Web of Science” with a more detailed investigation. Based on an additional individual
delayed but rapid jump in 2013 (see Figure 2). assessment the articles of Flack and Ross [25], Bazalo and
Moleta [26] and Bahrehmand et al. [27] were taken into
consideration for extended investigation.
With reference to the overall search strategy it is
imperative to point out its limits. The comprised search terms
were elaborated to provide a structured overview on the topic
but may not cover the research field in an exhaustive manner.
Thus, sources were replenished with an additional, more
random, search process that mainly focusses on tracing the
literature referenced by the identified results. However, a
general trend is obvious and the lack of investigations on the
overlapping of GD and layout planning in factories is shown.
In the following section, papers that were evaluated with
higher relevance for further research directions are presented.
Figure 2: Development of publications per year in Layout Planning and
Generative Design
3.2. Presentation of selected publications “Discover” [24]. Therefore, an overall workflow for GD
projects and key elements for the development of metrics as
Jo and Gero [19] developed a design model with well as a resuming data analysis are described. Due to its
evolutionary algorithms and focus on two aspects of major focus on office layouts, specific requirements of factory
relevance, namely the representation of the designer’s planning or manufacturing-environments were not
knowledge in a manner that can be accessed and evaluated investigated.
with evolutionary principles and the evaluation of the model’s The work of Flack and Ross [25] describes the adaption of
usefulness in distinct design problems. The described model a genetic algorithm for layout planning and compares
makes use of the so-called “evolutionary design based on the different approaches for handling multiple criteria. As a
genetic evolution system (EDGE)”, developed by Jo [28] and special feature, their approach allows the representation of
comprises a set rules that serve for the formalization and multiple floors and polygonal exteriors. Specific restrictions
interpretation of design information. In order to evaluate its of manufacturing are not investigated.
performance, the design process was tested in an office Bazalo and Moleta [26] investigate several algorithms for
planning case in comparison with an established space layout the exploration of architecture design alternatives in early
planning system by Liggett [29]. Thus, specific requirements planning stages. To do so, they describe general space
of a manufacturing environment are not addressed. In the planning approaches and the formalized space syntax theory
analysis, Jo and Gero highlight the generality for space (see Hillier et al. [30]) that provides the base for their
planning and the efficiency, i.e. the capability of hopping over algorithmic investigation of design alternatives combined with
large solution space in a short amount of time. physics simulation. Moreover, an approach for three-
Rosenman and Saunders [20, 21] introduced the self- dimensional planning in early planning stages is described and
regulatory hierarchical coevolution, an evolutionary algorithm aspects of optimization and necessary adjustments of the
for the application on complex design problems and especially algorithms are shown. Since the investigations focus on an
for so-called nonroutine designing, i.e. design tasks where early phase in building design, specific requirements of
relations between requirements and design alternatives are manufacturing environments are not considered.
hardly to determine. Basic characteristic of this algorithm is Bahrehmand et al. [27] present a layout solver which
the decomposition of complex problems into smaller problem considers several quality metrics as well as additional user
sets and the integration of coevolutionary principles. preferences in the development of design alternatives. The so-
Therefore, a relationship, or self-regulation, between these called “Interactive Layout Recommender System (ILRS)”
subcomponents is established by iterative evaluation of the generates an initial layout solution by means of an unspecified
distributed solutions regarding their contribution to an overall heuristic and improves the results in an evolutionary
target. The algorithm was tested in a space planning case for optimization algorithm combined with user ratings.
house layouts in order to proof that the generated layout Furthermore, the three-dimensional visualization is seen as an
structures satisfy previously defined requirements without elementary part for fine tuning by human planners. The
specific knowledge of the shapes and relations between the influence of user interactions and different algorithm features
single areas. Any relation to manufacturing-related problems are compared and the results substantiate the assumption of
is not described. decreased planning complexity by applying this hybrid
Fernando [22] proposes an approach that aims at the approach for general floor planning problems. However,
substitution of pencil sketches in early design stages. Since distinct characteristics of manufacturing or factory planning
the use of concurrent CAD programs in early design phases is are not addressed. By presenting related work and underlying
often assessed as inefficient and manual drawings are still concepts, Bahrehmand et al. emphasize that similar layout
preferred by designers. Therefore, he presents a prototyped planning solvers have primarily been applied to furniture
software application using blobs, defined as “[…] closed planning, residential space planning and urban design.
recursively subdivided curves […]”, as well as principles of The literature that has been derived with the presented
artificial life and genetic algorithms. Major aim is the search strategy reflects a lack of GD in factory layout
exploration of a multitude of design alternatives which would planning. Not least because the application of some
not be generated in a conventional, experience-based design evolutionary principle in factory layout decisions may not
process. The application was tested in a use case for spatial always be denoted with the selected search terms. Therefore,
planning without specific relation to any manufacturing- the additional indicated random search led to some findings
environment. that provide base for resuming research.
Nagy et al. [23] developed a graph-based model for the One example is an approach by Neugebauer et al. [31] who
evaluation of productive congestion. More specifically, this implemented a genetic algorithm for rough layout generation
so- called “Buzz Metric” quantifies of the desired in job-shop factories. Possible layout constellations are
accumulation of people in any architecture and may therefore derived from formalized properties of the components to be
serve as a (further) target criterion in generative design produced. However, the approach does not take boundaries of
projects. As highlighted by the authors, the underlying graph- subsequent fine planning (e.g. necessary aisles) or energy
based space syntax allows an efficient computation of results provision into account. Regarding the latter aspect, Jabi [32,
compared to extensive crowd simulations. Hence, a rather 33] developed the methodological framework as well as the
qualitative design factor is turned into a calculable form software library and implementation for coupled parametric
which still allows the investigation of a large solution space. design and energy simulation which may serve as a starting
The developed metric was tested in a two- dimensional office point for integrating energetic components into GD of factory
floor plan and furthermore applied by Nagy et al. with five layouts.
additional planning targets in a project called
4. Results for derived research concepts quickly calculable metrics, that reflect a desired aspect of the
factory layout, should be favored.
4.1. Challenges of Generative Design in factory planning Besides target definition, the boundaries restricting the
overall search space have further decisive impact on the
Requirements for a useful implementation of GD models output of a GD model and its variety [15]. In factory planning
are partially described in the presented articles (see section this is particularly relevant in brownfield planning cases when
3.2) and can additionally be derived from further building shapes, existing pillars or prevailing supply
mathematical investigations. More specifically, infrastructure vastly reduce freedom of planning alternatives.
interdependencies between the formulated solution space, the
user input or user interaction and the applied algorithm need 4.2. Framework of approach to the solution
to be considered. These requirements and their
interdependencies with factory planning are presented in the The presented challenges allude to the necessity of a
following.
structured approach when GD is considered as a value-adding
The solution space (also referred to as design space) should approach in factory layout design. A framework for this
be formulated by appropriate target criteria, parameters (i.e. approach is described in the following and additionally
decision variables) and constraints to provide a well-balanced depicted in Figure 3. The typical factory planning process is
solution set. Therefore, qualitative or intuitive requirements characterized by sequential, partly parallel, steps that may
and overall goals must be translated into quantifiable include some iterative decisions (see section 2.1). However, as
metrics [15]. According to Nagy, there are currently neither soon as distinct decisions in detailed planning (phase 3) are
general rules of thumb nor theoretical investigations for the made, changes may have complex and costly impact on
number of goals, variables and boundaries to generate a previous specifications. Therefore, it may be a promising
suitable mix between efficiency (continuity) and variety approach to work towards a frontloading or early-integration
(complexity) [24]. of dominant fine layout aspects into GD to observe their
Due to its interdisciplinarity, layout design requires factory impact from an early project phase onwards. As indicated,
planners to cope with a multitude of targets. As already typical and relevant examples of those aspects refer to the
mentioned, major aim is the reduction of material flow energy infrastructure. For instance, thermal losses and their
intensity or transport costs (see section 2.1). However, further effects in factories may correlate with conduit length, which
requirements of peripheral infrastructure will be more and could be translated into a further optimization criterion.
more integrated into early planning stages. A highly relevant This early adoption of fine layout aspects implies a well-
example is energy supply since manufacturing facilities are structured definition of overall requirements among the
increasingly required to cope with volatile energy supply project stakeholder in the preparation phase. Additionally,
which raises the amount of decentral energy sources, batteries data gathering and preparation gain even more relevance for
and complex grid structures (see e.g. [34]). Methodologies the definition of appropriate metrics. As already described,
that consider components of energy infrastructure in different target functions should be able to evaluate a requirement as
factory planning stages were, for instance, developed by precise as possible but their effect on efficiency of the search
Thiede [35] or Hopf [36], yet an integration of energy and process must be considered consistently. Finally, the selection
resource efficiency aspects into spatial optimization of and adaption of the appropriate algorithms and modeling tools
factories is still required. Additionally, various target criteria is relocated towards early planning phases when GD is
lead to further challenges when GD models are designed. As applied in factory layout planning.
generally known from multi-objective optimization, there are
multiple ways for handling multiple target criteria, like
weighted sum or Pareto optimization (see e.g. Marler and
Arora [37]). The latter is often applied in GD because of
visual interpretability and sufficient solution variety that is
found along the Pareto front [16] but as soon as the number of
target criteria moves beyond three-dimensional visuality,
analysis for human planners becomes more complicated.
Referring to the mathematical formulation of target criteria
an additional aspect regarding efficiency needs to be taken
into account. Since a criterion should reflect the impact of
different parameter settings as precise as possible, simulation
methods are generally conceivable as a target function for the
GD model. The already indicated linking of parametric design
and energy simulation by Jabi [32] provides base for
automated design evaluation. However, it is not unlikely that
GD algorithms operate with a population of 50 or more
individuals and more than 1000 iterations. Since simulation is
then required for each individual in every iteration, the
process becomes excessively time consuming. Thus, precise
but
Figure 3: Schematic representation of GD integration
5. Outlook [16] Stouffs R, Rafiq Y. Generative and evolutionary design exploration,
2015; 29, p. 329–331.
Based on the presented framework, several subsequent [17] Maciej Wypych, Matt Wash. Optioneering and Optimization Using
steps become conceivable. Firstly, the mathematical Generative Design and Evolutionary Solvers, Las Vegas; 2020.
[18] Jones EA. Is research necessary for equipment and layout planning?,
formulation and implementation of the proposed energy-
1955; 31, p. 903–906.
related aspects with currently available software environments [19] Jo JH, Gero JS. Space layout planning using an evolutionary approach,
should be conducted to encounter further hurdles that cannot 1998; 12, p. 149–162.
be elaborated from a methodological consideration. Secondly, [20] ROSENMAN M, Saunders R. Self-regulatory hierarchical coevolution,
developments in the available software tools and data 2003; 17, p. 273–285.
[21] Michael A. Rosenmann, Rob Saunders. Hierarchical co-evolution for
platforms are required to improve collaboration in a single
non-routine design, in Recent advances in intelligent systems and signal
“tool chain” from the very first draft onward. Especially processing; 2003, World Scientific and Engineering Academy and
developments in Building Information Modeling (BIM) Scociety, USA, p. 275–280.
already incorporate the described principles and serve as a [22] Fernando R. Space planning and preliminary design using artificial life,
base for ongoing research. Lastly, the implementation of in Rethinking Comprehensive Design: Speculative Counterculture -
Proceedings of the 19th International Conference on Computer-Aided
innovative layouts, apart from rectangular pillar structures,
Architectural Design Research in Asia, CAADRIA 2014; 2014, p. 657–
become tangible by GD principles in factory planning. 666.
[23] Nagy D, Villaggi L, Stoddart J, Benjamin D. The Buzz Metric: A Graph-
based Method for Quantifying Productive Congestion in Generative
Acknowledgement Space Planning for Architecture, 2017; 1, p. 186–195.
[24] Nagy D, Lau D, Locke J, Stoddart J et al. Project discover: An
application of generative design for architectural space planning, in
The authors gratefully acknowledge the financial support Symposium on Simulation for Architecture and Urban Design; 2017:
of the Kopernikus-project “SynErgie” by the Federal Ministry Toronto, Canada, 22-24 May 2017, Curran Associates, Inc, Red Hook,
of Education and Research (BMBF) and the project NY, p. 49–56.
supervision by the project management organization [25] Flack RWJ, Ross BJ. Evolution of Architectural Floor Plans, in
Applications of evolutionary computation; 2011, Springer, Berlin,
Projektträger Jülich (PtJ).
Heidelberg, p. 313–322.
[26] Bazalo F, Moleta TJ. RESPONSIVE ALGORITHMS An investigation
of computational processes in early stage design, in Proceedings of the
References 20th International Conference of the Association for Computer-Aided
Architectural Design Research in Asia CAADRIA 2015; 2015,
[1] Rist T. Ein Verfahren zur Modellierung von Produktionsbetrieben zum Association for Computer-Aided Architectural Design Research in Asia
Zwecke der Anordnungsplanung. Universität Stuttgart, Stuttgart; 2008. (CAADRIA), Hong Kong.
[2] Scheffer M. The forgotten half of scientific thinking. Proc Natl Acad [27] Bahrehmand A, Batard T, Marques R, Evans A et al. Optimizing layout
Sci U S A, 2014; 111, p. 6119. using spatial quality metrics and user preferences, 2017; 93, p. 25–38.
[3] Bundesamt für Justiz. Honorarordnung für Architekten und [28] Jo JH. A Computational Design Process Model Using a Genetic
Ingenieure: HOAI; 2013. Evolution Approach; 1993.
[4] VDI-Gesellschaft Produktion und Logistik. VDI 5200 Blatt 1 [29] Liggett RS, Mitchell WJ. Optimal space planning in practice, 1981; 13,
Fabrikplanung - Planungsvorgehen; 2011. p. 277–288.
[5] Wiendahl H-P, Reichardt J, Nyhuis P. Handbook factory planning [30] Hillier B, Leaman A, Stansall P, Bedford M. Space syntax, 1976; 3,
and design. Springer, Heidelberg; 2015. p. 147–185.
[6] Pawellek G. Ganzheitliche Fabrikplanung. Springer Berlin [31] Neugebauer R, Friedemann M, Riegel J, Wenzel K. Planning of job-shop
Heidelberg, Berlin, Heidelberg; 2014. factories leveraging genetic algorithms (Manuscript received March 30,
[7] Helbing KW. Handbuch Fabrikprojektierung. Springer 2009; revised July 30, 2009), in Manufacturing Research 2009.
Berlin Heidelberg, Berlin, Heidelberg; 2018. Conference proceedings; 2009, no pp.
[8] Schenk M, Müller E, Wirth S. Factory Planning Manual: Situation- [32] Wassim Jabi. Parametric Spatial Models for Energy Analysis in the
Driven Production Facility Planning. Springer-Verlag Berlin Heidelberg, Early Design Stages, in 5th annual Symposium on Simulation for
Berlin, Heidelberg; 2010. Architecture and Urban Design (SimAUD); 2014, p. 384–392.
[9] Michael Clauß. Methode zum Einsatz von Web 2.0-Werkzeugen in der [33] Jabi W. Linking design and simulation using non-manifold topology,
Fabrikplanung; 2013. 2016; 59, p. 323–334.
[10] Drira A, Pierreval H, Hajri-Gabouj S. Facility layout problems: A [34] Stoldt J, Prell B, Schlegel A, Putz M. Applications for models of
survey, 2007; 31, p. 255–267. renewable energy sources and energy storages in material flow
[11] Anjos MF, Vieira MVC. Mathematical optimization approaches for simulation, in WSC'18; 2018: Proceedings of the 2018 Winter
facility layout problems: The state-of-the-art and future research Simulation Conference, December 9-12, 2018, Gothenburg, Sweden,
directions, 2017; 261, p. 1–16. IEEE, Piscataway, Madison, p. 3287–3298.
[12] Kuhn HW. Variants of the hungarian method for assignment [35] Thiede S. Energy efficiency in manufacturing systems. Springer, Berlin;
problems, 1956; 3, p. 253–258. 2012.
[13] Banduka N, Mladineo M, Eric M. Designing a Layout Using Schmigalla [36] Hopf H. Methodik zur Fabriksystemmodellierung im Kontext von
Method Combined with Software Tool visTABLE, 2017; 16, p. 375– Energie- und Ressourceneffizienz, Wiesbaden; 2015.
385. [37] Marler RT, Arora JS. Survey of multi-objective optimization methods
[14] Lohmer J, Klausnitzer A, Lasch R. Advanced scientific algorithms for engineering, 2004; 26, p. 369–395.
in digital factory design applications, in Proceedings of the Hamburg
International Conference of Logistics (HICL); 2018, p. 56–57.
[15] Nagy D, Villaggi L, Benjamin D. Generative Urban Design: Integrating
Financial and Energy Goals for Automated Neighborhood Layout, in
Symposium on Simulation for Architecture and Urban Design
(SimAUD 2018); 2018: Delft, the Netherlands, 4-7, June 2018, The
Society for Modeling and Simulation International, Vista, California, p.
190–197.

You might also like