Modeling and Optimization of the Production and Logistic Concept of the Siemens "Factory One" in Zug

(1)

Modeling and Optimization of the Production and Logistic Concept of

the Siemens "Factory One" in Zug

Anika Splettstoesser

Master Thesis 16. September 2016

KTH Stockholm Prof. Dr. Daniel T. Semere

Prof. Dr. Amir Rashid

ETH Zürich

Prof. Dr. Konrad Wegener Simon Züst

(2)

(3)

Abstract

This thesis is part of a study to prove the feasibility of the logistics and production concept of a new production plant built by the business unit Building Infrastructure of the multicorporate enterprise Siemens in Zug, Switzerland. In order to perform this analysis the model of the factory built in the discrete-event simulation tool “Tecnomatix Plant Simulation” is used.

Prior student works aimed at answering the question whether the internal logistics should be performed by a milkrun or a flexible transportation system (FTS). They proved that the FTS is more robust. How- ever, their work did not fully prove the feasibility of the logistics concept and left two key questions unanswered. First, based on the high utilization of the FTS in the simulation the concern was raised whether the FTS could be the bottleneck of the system. Second, Siemens reduced the logistics area for the incoming and outgoing goods considerably and it needs to be determined whether the logistics area is large enough.

To answer the first question a new demand oriented control for the FTS is developed and several load scenarios are applied. The new control proves that the FTS is not the bottleneck of the system and can channel substantially more material through the system than anticipated in the production forecast. The sensitivity analysis has also shown that the FTS can operate at considerably lower speeds that expected.

In order to determine the space requirements for the logistics area all processes and material flows for this area are identified and implemented into the model. Based on the load scenarios and additional information it is determined that the provided area is too small. Furthermore, the simulation showed that the processes are sensitive to changes. Therefore, Siemens decided in a workshop based on the simulation to enlarge the area to provide additional capacity. Finally, it was decided to increase the area by 50%.

By answering these two questions this thesis finalizes the feasibility analysis of the logistics concept.

Furthermore, this work lays the foundation for analyzing the feasibility of the production concept by providing a detailed investigation into the new production concept of Siemens and by proposing a model for the production cells which complements the logistics simulation.

(4)

(5)

Sammanfattning

Denna rapport är en del av en studie för att säkerställa planeringen av ett logistik- och produktionskoncept för en ny produktionsanläggning som byggts av affärsenheten Building infrastruktur av företaget Siemens i Zug, Schweiz. För att utföra denna analys har fabriken avbildats med hjälp av en simuler- ingsmodell i programvaran “Tecnomatix Plant Simulation”, baserat på diskret simulering.

Tidigare studentarbeten har fokuserat på att besvara frågan om den interna logistiken bör utföras med hjälp av milkrun eller ett flexibelt transportsystem (FTS). Efter applicering av ett flertal scenarier i simu- leringsmodellen har det bevisats att FTS är robustare än milkrun och därför har Siemens beslutat sig för att använda FTS. Tidigare arbeten har dock inte undersöktgenomförandet av logistikkonceptet och har därmed lämnat två viktiga frågor obesvarade. Den första frågan gäller den höga utnyttjandegraden av FTS, vilket gör det potentiellt möjligt att FTS kan vara flaskhalsen i systemet. I och med att FTS är en stödprocess är det viktigt att denna inte är flaskhalsen i de värdeskapande tillverkningsprocesserna.

Den andra frågan gäller logistikområdet för inkommande och utgående gods som Siemens betydligt har minskat betydligt. Här måste det fastställas huruvida logistiken kan funktionera i det mindre utrymmet eller om området måste utvidgas.

För att besvara den första frågan utvecklas en ny FTS styrning där flera lastscenarier tillämpas. Den nya styrningen bevisar att FTS inte är flaskhalsen i systemet och kan hantera betydligt mer materialflöden än förväntat. Sensibilitetsanalysen har även visat att FTS kan arbeta i betydligt lägre hastigheter än väntat.

För att bestämma ytan för logistikområdet måste alla processer och materialflöden för detta område identifieras. Baserat på lastscenarier och ytterligare information kan det identifieras att det planerade området är för litet. Dessutom visar simuleringen att processerna är känsliga för förändringar. Därför beslutade Siemens under en workshop att utvidga detta områd för att klara ytterligare kapacitet. Det beslutades att öka arean med 50 %.

Genom att svara på dessa två frågor säkerställer detta arbete genomförbarheten av det planerade logistikkonceptet. Vidare lägger detta arbete en grund för att analysera genomförbarheten av produktionskoncept genom att tillhandahålla en detaljerad undersökning av det nya produktionskoncept för Siemens och genom att föreslå en modell för produktionsceller som kompletterar logistiksimulering.

(6)

(7)

Zusammenfassung

Diese Masterarbeit ist Teil einer Studie, die von der Business Unit Building Technologies von Siemens in Auftrag gegeben wurde, um das Produktions- und Logistikkonzept ihrer neuen Fabrik in Zug in der Schweiz im Vorfeld zu überprüfen. Um diese Analyse durchzuführen wird ein Modell der Fabrik in der ereignisorientierten Simulationssoftware “Tecnomatix Plant Simulation” verwendet.

Die Vorgängerarbeiten sind der Frage nachgegangen, ob die interne Logistik durch einen Milkrun oder ein flexibles Transportsystem abgewickelt werden soll. Unter Anwendung mehrerer Szenarien in der Simulation haben sie nachgewiesen, dass das FTS robuster ist. Allerdings konnte ihre Arbeit die Durch- führbarkeit des gesamten Logistikkonzepts nicht final nachweisen, da zwei zentrale Fragen unbeant- wortet blieben. Zum einen wurden Bedenken geäussert, dass das FTS aufgrund seiner hohen Aus- lastung zum Engpass im System werden könnte. Zum anderen hat Siemens den Logistikbereich für den Wareneingang und Warenausgang in der neuen Fabrik erheblich reduziert. Dementsprechend ist nachzuweisen, ob alle Prozesse und Materialflüsse in dem kleineren Bereich abgewickelt werden kön- nen.

Um die erste Frage zu beantworten, ist die Steuerung des FTS anhand von definierten Kriterien neu aus- gerichtet worden. Im Anschluss wurde diese anhand verschiedener Szenarien getestet. Hierbei erweist sich die neue Steuerung als wesentlich effizienter und ist nicht länger als potentieller Engpass anzusehen, weil erheblich mehr Material durch das FTS geschleust werden kann, als Siemens in seinem Produk- tionsforecast antizipiert. Um zu ermitteln, wie viel Platz der neue Logistikbereich benötigt, wurden alle Materialflüsse und Prozesse identifiziert und die fehlenden in die Simulation integriert. Die Ergebnisse der Simulationsläufe zeigen, dass der Logistikbereich zu klein geplant wurde. Als Konsequenz auf diese Analyse hat Siemens in einem Workshop entschlossen den Bereich um 50 % zu vergrössern und dadurch Kapazitätspuffer zusätzlich zu den Simulationsergebnissen zu berücksichtigen.

Durch die Beantwortung dieser Fragen ist die Durchführbarkeit des Logistikkonzepts erwiesen. Zusät- zlich zur Vervollständigung der Analyse der Logistik legt diese Arbeit die Grundlage für die Analyse des Produktionskonzepts durch die detaillierte Betrachtung des Konzept und die Entwicklung eines Vorschlags zur Implementierung der Produktionsinseln als Erweiterung des Logistikmodells.

(8)

(9)

List of Figures

1.1 Rendering of the Factory One by Siemens . . . 2

1.2 Sketch of the Factory One . . . 2

1.3 Prior work . . . 3

2.1 Factory types from customer perspective . . . 8

2.2 Competitive factors . . . 9

2.3 Connected value flow . . . 9

2.4 Product process matrix . . . 10

2.5 Cellular manufacturing . . . 10

2.6 Scheme for decision methods and descriptive models . . . 11

2.7 Material handling system . . . 13

2.8 Core and sub-processes of industrial production . . . 14

2.9 Transportation curve . . . 14

2.10 Example layout of a warehouse . . . 15

3.1 Concept of the FTS control . . . 20

3.2 Old model of the FTS in plant simulation . . . 21

3.3 Attribute manager for the entry and exit of the boxes . . . 21

3.4 List for arriving boxes . . . 22

3.5 Parameter setup for the stations in the FTS . . . 23

3.6 Routine for pausing the lifter . . . 23

3.7 Selecting the next target . . . 23

3.8 Representation of floors in the control . . . 24

3.9 Halting the lift when next target is reached . . . 24

3.10 Request for loading and unloading . . . 24

3.11 Unloading procedure . . . 25

3.12 Procedure for loading the boxes is triggered . . . 26

3.13 New model of the FTS in plant simulation . . . 27

(14)

List of Figures

3.14 Sankey diagram in the FTS . . . 28

3.15 Implications of different lifter velocities . . . 29

3.16 Number of items at entrance and exit into the FTS at 8 m/s . . . 30

3.17 Number of items at entrance and exit into the FTS at 0.35 m/s . . . 30

4.1 Current layout logistic . . . 36

4.2 Future layout logistic . . . 38

4.3 Goods receiving area . . . 41

4.4 Shipping area . . . 41

4.5 Model of logistics area . . . 42

4.6 Logic of the source networks . . . 43

4.7 Logic of the palletizing networks . . . 44

4.8 Model of logistic within logistics area . . . 45

4.9 Transfer area for 3PL . . . 46

4.10 Model of material handover networks . . . 47

4.11 Decision tree handover of material from logistics . . . 48

4.12 Logic of depalletizing robot . . . 48

4.13 Logic of palletizing robot . . . 49

4.14 Drain network of 3PL . . . 49

4.15 Results of multiple scenarios, interpretation in Table 4.4 . . . 50

4.16 Rough layout based on simulation results . . . 53

4.17 Utilization of logistic workers at standard volume . . . 54

4.18 Utilization of logistic workers at 25% volume increase . . . 55

4.19 Utilization of logistic workers at different arrival times . . . 56

4.20 Utilization of logistic workers including one additional worker . . . 57

4.21 Unloading time 3PL at standard volume . . . 57

4.22 Unloading time 3PL at 25% volume increase . . . 58

4.23 Unloading time 3PL at different arrival times . . . 58

4.24 Occupancy bulky items in the temporary storage at standard volume . . . 58

4.25 Occupancy bulky items in the temporary storage at at 25% volume increase . . . 59

4.26 Occupancy bulky items in the temporary storage at at different arrival times . . . 59

4.27 Unloading time 3PL at standard volume with extended capacity for bulky items in the temporary storage . . . 59

4.28 Unloading time 3PL at 25% volume increase with extended capacity for bulky items in the temporary storage . . . 60

4.29 Unloading time 3PL at different arrival times with extended capacity for bulky items in the temporary storage . . . 60

4.30 Occupancy bulky items in the temporary storage with extended capacity at standard volume 60 4.31 Occupancy bulky items in the temporary storage with extended capacity at at 25% volume increase . . . 61

4.32 Occupancy bulky items in the temporary storage with extended capacity at at different arrival times . . . 61

4.33 Unloading time 3PL at standard volume with extended capacity for bulky items in the temporary storage and extra worker . . . 61

4.34 Unloading time 3PL at 25% volume increase with extended capacity for bulky items in the temporary storage and extra worker . . . 62

4.35 Unloading time 3PL at different arrival times with extended capacity for bulky items in the temporary storage and extra worker . . . 62

4.36 Occupancy bulky items in the temporary storage at at 25% volume increase . . . 62

xii

(15)

List of Figures

4.37 Bulky items in the temporary storage at the end of the simulation in the volume increase

scenario . . . 62

4.38 Occupancy bulky items in the temporary storage with additional shipping at 25% volume increase . . . 63

4.39 Bulky items in the temporary storage at the end of the simulation in the volume increase scenario . . . 63

4.40 Transitions between push and pull in the simulation . . . 63

5.1 Product process matrix . . . 68

5.2 First sketch of possible future production layout for the ground floor . . . 68

5.3 First sketch of possible future production layout for the first floor . . . 68

5.4 General concept for production cells . . . 71

5.5 Model of SMD and THD souldering . . . 71

5.6 Model of CEVA production cells . . . 72

5.7 Model of GAP production cells . . . 72

(16)

(17)

List of Tables

2.1 Examples for simulation as an optimization tool . . . 12

3.1 Current performance of FTS lifter . . . 18

3.2 Current arrival schedule for the trucks at Factory One . . . 19

3.3 Current arrival schedule for the trucks at Factory One . . . 26

3.4 Moved items by old and new control . . . 27

3.5 Moved items between the floors . . . 28

3.6 Comparison between calculated distance and traveled distance in the simulation . . . 28

3.7 Performance of the new control under different load scenarios . . . 31

3.8 Performance of the old control . . . 32

3.9 Performance of the new control . . . 32

3.10 Comparison of performance of the new control with the targets . . . 33

4.1 Space requirements in square meter . . . 37

4.2 Design of experiments for logistics area . . . 39

4.3 Delivery schedule for parcels . . . 40

4.4 Calculated space requirements from the results of the variation . . . 51

4.5 Space requirements based on the simulation . . . 52

4.6 Final results space requirements . . . 64

(18)

(19)

1

Introduction

1.1 Motivation

Siemens is a multinational group with around 350’000 employees worldwide and is organized into ten business units. The headquarter of one of these units is in Zug, Switzerland. For this business unit

"Building Technologies" 30’000 employees worldwide develop, manufacture and sell products for building automation, safety and control such as fire detectors. The production sites are spread worldwide, but the main plant as well as its R&D is located in Zug.

To remain competitive within the worldwide manufacturing network Siemens currently builds a new plant in Zug, which is called "Factory One". A rendering of the new factory can be found in Figure 1.1.

The building project is the second step of the production consolidation of the business unit "Building So- lution". The project does not only include the factory but also an office building for the other functions.

In the first step another Swiss production facility was moved into the two existing manufacturing build- ings in the city center of Zug. In the second step the production in these two building is consolidated in the new building of the "Factory One". The investment into the production facility alone is approximately CHF 120 - 150 million (1’040 - 1’300 mio. SEK) and the start of production is in 2019. Due to the high investment a feasibility analysis of the logistics and production concept shall be conducted beforehand.

This thesis uses a simulation model of the future factory to contribute to the feasibility analysis.

1.2 Factory One

In the following some general information about the setup of the Factory One is presented (cf. Figure 1.2). In the building the production stretches across three floors. The production is located on the ground floor and the first floor the whilst the basement is used as a storage. It is called a temporary storage as apart from the safety stock the material is only stored for a few hours before it is sent to production. The temporary storage is connected to the production in two ways. The first way is through lifts for the bulky

(20)

1 Introduction

Figure 1.1: Rendering of the Factory One by Siemens

items which are transported manually with forklifts. The second way is through a flexible transportation system (FTS) for the standardized boxes. The FTS is an automatic lift system with three parallel lifts along the length of the building which is connected to an automated warehouse within the temporary storage. In addition, a (de)palletizing robot handles the boxes between the FTS and the logistics area on the ground floor.

The FTS can transport the boxes between the floors and through the temporary storage from one lift to another. The underlying intention of the logistics concept is that most of the material should be transported in boxes and use automation as much as possible. The transportation from the lifts to the production is done manually by logistics workers. Another task of these workers is to handle the incoming and outgoing material flow in the logistics area. The incoming production material is solely delivered by a third party logistics provider (3PL) called Senn. The 3PL receives the shipment from the suppliers, per- forms the incoming inspection, packs the material into the standardized boxes and delivers the material several times per day to the factory. It also receives the outgoing material flow with the exception of the finished goods. Those are shipped to the distribution center (DC) in Germany from where the products are distributed to the customer.

Figure 1.2: Sketch of the Factory One

2

(21)

1.3 Related Work

The most important decision taken since the start of the project is to offshore one part of their production to other Siemens plants at the same time as the move of the production from the second building into the other old building during the summer 2016. The older building will then be demolished to make space for the new factory. Furthermore, the quotations and thus the exact specifications for the FTS, the temporary storage and the (de)palletizing robot are pending.

1.3 Related Work

In each of the two semesters prior to this thesis a pair of students worked on the Factory One in cooper- ation with the consultancy Acel & Partner. They laid the foundation by building a model of the Factory One to answer the question whether the logistics within the factory should be handled by an FTS or by a milkrun. A milkrun delivers material according to a schedule to production by transporting it on multiple wagons which are drawn by steered towing vehicle. In both student works the FTS proved more efficient and Siemens opted for the FTS. A short overview of their studies can be found in Table 1.3. This thesis builds upon their work and addresses the open topics such as a better control of the FTS and a more detailed logistics area. The focus of this thesis is discussed in more details in 1.4.

Students Task Results Open Topics

Stefan Hutzli Felix Winkler (FS 2015)

Modeling and evaluation of two logistic concepts (milkrun vs. FTS) at the Factory One

Simulation model of

“Factory One” with focus on logistics FTS seems to more robust than milkrun

Further evaluation based on more detailed model needed to take investment decision (milkrun vs. FTS) Patrick Vettiger

Matteo Odermatt (HS 2015)

Modeling and evaluation of the variants milkrun &

FTS for production and logistics in the Siemens Factory One

Update of the simulation

§ New building layout

§ More detailed &

realistic logistic system Proof: FTS more reliable than milkrun (investment decision taken by Siemens)

Better steering depalletizing robot &

FTS

More details in the material flow (especially pallets, recycling, two-way packaging) detailed production cells

Figure 1.3: Prior work of [8, 16, 24, 30]

1.4 Problem Formulation

Based on the prior work and the current status of the Factory One project the problem statements are formulated for this thesis. The ultimate goal of the entire study is to probe the feasibility of the logistics and production concept. However, this work is too extensive to finish it with this thesis. Therefore, limits of this thesis are set. At the end of this report further steps for following student work are framed.

Logistics

The feasibility of the entire logistics concept needs to be proven before the production concept can be implemented into the model in detail. So far, the FTS has been found to be more efficient than the

(22)

1 Introduction

milkrun. However, there are reasonable doubts whether the FTS might be the bottleneck of the system and whether the logistics area around the truck ramps is large enough.

With respect to the FTS the logic in the automated warehouse needs to be improved. Currently, no analysis has been performed to determine the impact of the speed of the transportation of the boxes to and from their storage location in the FTS lifts. This is crucial as lighter boxes can only be lifted with reduced speed as the acceleration forces of the lifter could dislocate the box. However, the weight of each box is not instrumental for the storage as the shelves can accommodate all combinations of possible box weights.

The second and more crucial issues with the FTS is the control of the vertical lifters in the lifts which move the boxes between the different floors. The current status of the model works on a FIFO basis without any priority or optimization. Furthermore, it reflects reality imperfectly as the lifts always move up to the second floor first even if the boxes just needs to be transferred one floor up or down. Therefore, the most important task is to develop a control which optimizes traveling time whilst taking all boundary conditions into account.

Key question: Can the FTS handle the transportation of all boxes to arrive at the right time at the production cells without unnecessary movements?

The logistics area for incoming and outgoing material flow is only roughly implemented. In this area the incoming materials from the 3PL needs to be moved from the truck into the area and the boxes need to be taken from the pallets into the FTS by the (de)palletizing robot. However, this is not the only incoming material flow. In addition, bulky items are also brought into the system by the 3PL and material arrives by post or parcel delivery. All materials which come into the system need to leave the system in the same area. However, at the moment there is no clarity about the exact flows in and out of the system. Furthermore, the current model does not reflect the material flow as a closed loop system.

Instead material such as packaging and pallets are eliminated in the model at certain points. In order to successfully plan the area these material flows need to be fully analyzed. Based on this analysis a layout of the area needs to be derived and tested in the simulation model. The most important question to be answered is whether the currently planned space for the area is sufficient. As this information is vital for the further planning of the new factory it needs to be derived first. Furthermore, the information about this ’big cell’ behind the truck ramps and the material loop inside the factory is vital for the design of the size of the production cells and their connection to the logistics system.

Key question: Is it possible to channel the future material flow through the planned logistics area?

If these two question have been answered positively the complete logistics concept is verified and the earlier work would be concluded. Afterwards the production concept can be implemented into a fully functioning logistics model.

Production

The future production concept needs to be verified as the new building presented an opportunity to change it substantially. In this thesis the production is analyzed and a proposal for evaluating possible layouts is given. Based on this the concept for modeling the production is developed. As mentioned earlier this thesis is the first step in evaluating the feasibility of the production concept.

Key question: How can production concept be analyzed and modeled to prove that the future factory can deliver the required output?

Based on the problem statements the project can be divided into three sub tasks: the FTS, the area for

4

(23)

1.5 Plant Simulation

incoming and outgoing goods as well as the model of the production. However, these problems are highly related as together they represent the value adding process of the factory. These links are further discussed in the next chapter.

1.5 Plant Simulation

For the simulation the eleventh version of the software Tecnomatix Plant Simulation from Siemens is used. It is a discrete-event simulation tool focused on logistics systems including production. [21] The program is used to ensure continuity as earlier works have also used this software.

(24)

(25)

2

Frame of Reference

In this chapter the state of the art is presented. First, the prodution concept of the Factory One is described. Within this section the focus is set on finding the most important parameters for the new factory in order to evaluate the concept. Second, the different sub-areas such as the logistic area, the flexible transportation system (FTS) and the production cells are analyzed in more details. The state of the art is investigated for these in order to find feasible approaches to answer the key questions presented in Chapter 1.4.

2.1 Operations Strategy

In order to understand the key parameters for the project the Factory One is put into context by looking at the different factory types from the customer perspective. Ultimately, the customer will decide on the success of and therefore on the survival of the plant. Even though the central distribution center is an internal customer this view is essential as the Siemens factory in Zug is in direct competition with all Siemens plant worldwide and potential suppliers. From Wiendahl et al. [28] the following factory types can be distinguished as shown in Figure 2.1:

• high tech factory

• responsive factory

• breathing factory

• customer individual factory

• variant flexible factory

• low-cost factory

Based on the criteria stated in Figure 2.1 and statements from Siemens the Factory One can be regarded

(26)

2 Frame of Reference

Figure 2.1: Factory types from customer perspective [28]

as a mix of high tech and variant flexible factory for the following reason. First, Siemens in Zug is the competence center and global headquarter for the segment Building Technology. This means that the plant is responsible for the introduction of new products into the market and into production. Therefore, it has to be ahead of all other Siemens production sites in terms of technology. In addition, some of the new products are transferred to other plants after a successful market launch and the production methods have been sophisticated. Second, a broad range of product families and variants are produced in Zug with varying volume. Due to strategic reasons some products are produced in very low quantities.

In order to derive the key factors, the competitive factor matrix for each factory type proposed by Wien- dahl et al. [28] is used to analyze the Factory One. Based on the two types, high tech and variant flexible factory, the result for Siemens is shown in Figure 2.2. It can be said that the Factory One should have a focus on quality and flexibility while achieving a high delivery reliability and a competitive cost structure. Flexibility is used as broader term for innovativeness, learning speed and changeability. These factors are important for Siemens as the Future Factory can only compete based in a high cost country if it is able to adapt to changes in product mix and portfolio quickly and without raising cost.

To achieve these competitive factors the entire value-adding process in the factory needs to be optimized.

Even though the value is added to the product in manufacturing, the plant logistics need to be optimized as well. This optimization shall support production whilst using a minimum of resources. This view on the value chain stated by ten Hompel & Schmidt [22] is transferred to the Factory One. As a result its connected value chain is shown in Figure 2.3. In addition, this approach can be used to define the system boudaries for the simulation, which are indicated by the black box in Figure 2.3. As a consequence, this thesis shall look into the connected value flow into the logistic area (incoming and outgoing goods), the FTS and the production. Therefore, these area are investigated further in this chapter.

8

(27)

2.2 Production

Figure 2.2: Competitive factors [28]

Figure 2.3: Connected value flow in the Factory One

2.2 Production

First of all, the type of production is determined using the product process matrix described in Hill &

Hill [7] based on Hayes & Wheelwright [6]. In this matrix the production volume per product is matched with the process choice ranging from project type to continuous production. The optimum for process choice depending on the volume according to literature is shown in Figure 2.4 as a dotted line. In the current state of the Siemens production some lines like SMD (surface mounted device, assembly process circuit boards) feature a functional/ job shop layout while the final assembly is grouped into production cells. The current state is indicated as a black bubble in Figure 2.4. Between the different lines the material is transported in batches from one process to the next. This layout is not ideal for the higher volume products and induce a lot of inventories within the factory. Therefore, in the future state the production cells are connected into a flow and in greater proximity. According to the matrix proposed by Hayes & Wheelwright [6], Hill & Hill [7] this is a feasible course of action. However, one has to bear in mind that the model only gives a rough classification as it is not clearly stated how much units a low or high volume represents. The classification of different production lines in the literature is based on examples from the industry. In Hill & Hill [7] the example for a line would be automotive production.

Despite having high volume lines such as SMD the Siemens production volume in Zug is far lower than the volume associated with automotive production. Therefore, the Factory One needs to implement a production layout which is an improvement to their current batch-and-queue significantly reducing its inventory without decreasing its flexibility by setting up a line production for all its processes.

According to Wang [26], Torabi & Amiri [23] cellular manufacturing could be the answer to this problem.

In cellular manufacturing workstations and equipment are grouped together in order to produce a produce in one-piece-flow. It can be used to produce a high variety of products. Especially, it is feasible if the

(28)

Figure 2.4: Product process matrix (based on [6, 7])

demand is stable. It is advised to use a functional layout for fluctuating demand. Figure 2.5 is an example for a cellular manufacturing layout. The machines and stations are set up in a U-shape while the workers move in an inner circle possibly taking the material with them. It can be derived that the maximum number of workers in the system depends on the number of stations.

Figure 2.5: Example of cellular manufacturing [4]

Based on the discussion with Siemens they assume that the demand will remain as stable as it has been in the past. Their main challenge is to produce a high variety of production. In addition, the one-piece-flow within the cells reduces the inventory considerably. The FTS could then be used to transport significantly smaller batches between the cells without a considerable increase in labor as the cells can be grouped closely together in the proximity of the next lift of the FTS.

Therefore, the optimization of a cellular manufacturing system can be divided into two sub problems:

the cell design and the optimization within the entire system. Therein, according to literature the cell design problem can be further divided intoo three to four steps, which are either done in parallel or consecutively. (cf. Wang [26], Torabi & Amiri [23], Javadi et al. [9], Salum [19]

10

(29)

2.2 Production

1. cell formation 2. intra-cell layout 3. inter-cell layout 4. (evaluation)

As the cell design is the basis for the optimization of the entire system as targeted by Siemens the state of the art for this problem is discussed in more details in the following paragraphs. Even though the literature fairly agrees on the necessary steps there is no agreement on the approach how to solve it. Some sources only deal with how to form the cells. For example King [10] proposed a part-machine-matrix in which parts which used the same machines could be grouped together. Other approaches include using heuristics [31] or fuzzy math [23]. Javadi et al. [9] proposes a two steps non-linear optimization problem. In the first step the cell formation is solved before computing the intra-cell and inter-cell layout simultaneously. Whilst Wang [26] structures the same problem steps by the identification of part families, system design and heuristics. There seems to be an common understanding as stated by Askin [2] that the final step of evaluating the found solution is missing. This is problematic as it cannot be said whether a solution supports the factory planning in the best way, or even a better way, than another. This problem is addressed by Salum [19]. In this paper the lead time is first improved by a simulation the entire sytem. In the second phase machines are then grouped closely to each other which have shown similar parameters in the simulation which indicate mimimum total material lead time. Even though Salum [19] only focuses on one specific parameter and the simulation is used to understand the system, Wang [26] pointed out that the simulation of a model can be used to evaluate the proposed solution.

In order to successfully rank potential solution clear success parameter (or key performance paramters) should be defined as implemented by Salum [19]. After a cell design has been found in the next step the entire system including e.g. logistics needs to be optimized. [13] provides a comprehensive overview on this optimization as shown in Figure 2.6. Based on the requirements of Siemens to cater for changes in parameters and a dynamic development the discrete event simulation is discussed in more details.

Figure 2.6: Scheme for decision methods and descriptive models [13]

(30)

Table 2.1 represents an overview of different studies of the application of discrete event simulations.

These confirm the assumption that simulation represents an appropriate tool for answering the key questions for the Factory One projects. For example, Williams & Ülgen [29] state based on simulation applications in the automotive industry that simulation can be used for testing or evaluation plans or new concepts with high capital expenditure without high expenses for the evaluation. This is true for the Siemens case as well. Its new factory cost approximately CHF 120-150 million. Any occurring cost for setting up the simulation are fractional compared to this investment. Furthermore, according to Bangsow [3] process simulation can be used to safeguard an entirely digitally planned manufacturing process. This can statement can be applied to Siemens as well. The new logistic process has not yet been tested in reality. However, in order to ramp up the production in the new building as smoothly as possible the logistic processes need to function properly. Voorhorst et al. [25] and Kulkarni & Gowda [11] both discuss the advantages of simulation to find and evaluate feasible layouts for production. A necessity for the success of the project as discussed before. As a result, based on these studies it can be derived that discrete event simulation is an appropriate tool to test and evaluate the novel production concept for the Factory One.

Case Findings Parallels/ Comment

Simulation applications in the automotive industry [29]

Simulation is perfect for testing/ evaluating plans w/o capital expenditure for capital intense projects;

Good to test new concepts in manufacturing, logistics, etc.

Parallels for the structure and the data collection;

Presents two successful examples

Coupling digital planning and discrete event simulation, taking the example of an automated car body in white production [3]

Process simulation safeguarding and entirely digitally planned manufacturing process

Good discussion of steps to create simulation;

Similar to the framework presented in the lecture "plant simulation" []

Optimizing a highly flexible shoe production plant using simulation [25]

Discussion of optimal layout;

Measured variables: overall performance, labor utilization, bottlenecks (queues);

Result: evaluation which product mix can be manufactured

Variability similar to Siemens

Validating the existing solar cell manufacturing plant layout and proposing an alter- native using plant simulation [11]

Powerful tool to evaluate layout propos- als;

Process improvement in order to raise machine, human and system performance and identify bottlenecks;

KPI: cycle time reduction, productivity increase, reduction in traveling time, resource utilization factor

Good orientation;

Also cost analysis

Table 2.1: Examples for simulation as an optimization tool

2.3 Flexible Transportation System

Neumann [15] describe a similar problem to the FTS at Siemens. In their study a complex automated material handling system forms the interface and link between a warehouse, two production areas and an order picking area. A sketch of the system is shown in Figure 2.7. This is very similar to the temporary storage area, the two production floors and the logistic area in the Factory One. In both cases the system handles numerous crossing flows of palletized raw materials, products, packaging material etc.. In the paper simulation was used to find the limits of the system. As a result the author could identify the important parameters of the system. First, the all pallets flows through the system needs to be identified.

Second, the process control needs to be defined. This control includes the priority of different flows at crossing points, the principle of scheduling (in this case FIFO) and how to deal with empty positions. In

12

(31)

2.4 Logistic Area

their study they particularly encountered problems with manual sinks and depalletizing robots.

Figure 2.7: Material Handling System [15] in [12]

With all these characteristics the study represents a nearly perfect parallel to the Siemens case starting from the flow setup to the depalletizing robot. Therefore, the two discovered parameters should also be applied to the Factory One. As the first step the flows in the system need to be defined. This has already been done in earlier students’ work. Therefore, the task in this thesis is define a more efficient process control which consist of a priority of flows and scheduling mechanism. For the priority a decision diagram as used by Neumann [15] could be applied. Besides FIFO other mechanism for scheduling from production planning described by Nahmias [14] e.g. could be used such as:

• FSFS: First Come First Served

• SPT: Shortest Processing Time → minimizes total completion time, mean flow time and total lateness

• EDD: Earliest Due Date → minimizes maximum lateness and maximum tardiness

• CR: Critical Ratio → time remaining before due date/ remaining processing time → the smallest CR goes first

• S/RO: mimimum slack time per operation → time remaining before due date - remaining processing time

2.4 Logistic Area

The Logistics area hosts several important activities (vital for both performance and cost), such as: receiving, depalletizing, put-away (as an interface to the warehouse), packing and loading.[18] These processes are part of the overall manufacturing process as shown in Figure 2.8.

(32)

Figure 2.8: Core and sub-processes of industrial production [28]

In order to avoid bottlenecks the area should not reach its limits in transportation and storage. Its limits are set by insufficient capacity (either in the processes such as the speed of the handling system or insufficient space for material storage and handling for the required output) and system dynamics. [28]

These limitations are shown in Figure 2.9.

Figure 2.9: Transportation and storage curve [28]

These limits can be derived by queuing theory as discussed by Ozaki et al. [17] or simulation as presented by Emami et al. [5] depending on the boundary parameters. As a result the minimum space requirements for the warehouse can be acquired and a layout of the area can be derived. An example of such a study is shown in Figure 2.10.

14

(33)

2.4 Logistic Area

Figure 2.10: Example layout of a warehouse [20]

(34)

(35)

3

Control of the FTS

To structure the approach to answer the question, whether the FTS can transport the boxes to arrive at the right time at the production cells without unnecessary movement, the methodology presented by Acel [1]

in the lecture for plant simulation is used. The methodology represents a thorough approach to structure a simulation project. It is also applied to the logistic area and the production cells.

3.1 Situation Analysis

All earlier studies pointed out that the control of the lifter in the FTS needs to be improved (c.f. Section 1.3). The lifter in the FTS moves the boxes from one level to another. In total the system is connected to three levels: basement, ground floor and the first floor. The three automated lifts operate in parallel in the factory. The lifter can transport up to two boxes, one on the right side and one on the left side.

The boxes are put into the system from the left or the right entrance by the logistics workers. Until they reach the lifter the feeding of the boxes on belts follows the FIFO principle. Currently the lifter moves in circles from the cellar to the second floor and back without any optimization of its movement. Boxes are put automatically on the lifter if it moves past the level of the box and the right side is empty. The boxes are disembarked when the lifter passes their target level. The first lifter is under the highest load as all boxes from and to the palletizing robot are moved by it. Therefore, it is predestined to be the bottleneck once the volume increases.

In order to avoid unnecessary movements the control needs to be changed. These unproductive movements between floors lead to higher utilization of the system and limits the output of the FTS substantially. Furthermore, the improvement needs to include the system how the lifter moves to the target floor but also needs to consider by which logic the boxes are selected to be transported. In the current system there is no priority. Boxes will be selected when the lifter passes them. This can lead to some boxes waiting for a long time.

Furthermore, all arriving boxes from logistics and production are first transported to the temporary stor-

(36)

3 Control of the FTS

age in the basement. This logic shall remain for the following reasons. First, the material shall only be transported to the production or logistics area if a cell or the palletizing robot calls it. Following the pull principle is essential for control the production efficiently. In addition, the space at the production cells is limited due to space constraints and the production philosophy (cf. lean production). Therefore, the material is delivered to the production cells when needed and the material reach is reduced in comparison to the current production concept. Second, the temporary storage is more flexible in control the material flow as it can operate independently from the workers through its automated picking system. Third, the boxes do not necessarily arrive in the lift which is closest to their destination. Therefore, the temporary storage in the basement is used to sort the material. This procedure is cheaper and more efficient than letting the logistic workers carry material for a longer distance. This logic can be improved once the system can differentiate the called material. e.g. by an attribute. As a consequence, the control should be flexible enough to accommodate this change. Finally, the new control is tested against the key question whether the FTS can transport all boxes to arrive at the right time at the production cells without unnecessary movement.

3.2 Target Definition

Based on the discussion in Section 3.1 the performance of the FTS needs to be improved. Its current performance is shown in Table 3.1. Based on the utilization the lifter 1 can be confirmed as a potential bottleneck. Its utilization is the highest as it is connected at the ground floor to the (de)palletizing robot.

Therefore, all boxes enter and leave the FTS through lifter 1. In order to eliminate it as a bottleneck the following targets are set. First, the traveled distance has to be cut by the unproductive movements between levels. Second, the reduced distance should also have an impact on utilization. In order for the system to work at an increased volume the utilization should be low at normal volume. Third, the relative occupancy should increase as the lifter should transport two boxes as often as possible. In an unfeasible solution for the control the number of moved objects would decrease. This behavior would clearly indicate that the FTS is underperforming and the entire system would be instable. In summary the following targets shall be achieved:

1. reduce utilization (working time) of first lifter by at least 50%

2. cut distance traveled of the first lifter by at least half 3. improve relative occupancy by at least 20%

4. transport at as least as many objects as with the current control

Lifter 1 Lifter 2 Lifter 3

Utilization 30% 7% 8%

Relative occupancy 22% 6% 4%

Distance traveled 1’066 km 273 km 290 km

Moved objects 34’095 15’182 11’112

Table 3.1: Current performance of FTS lifter

18

(37)

3.3 Problem Definition

The task is to develop a control for the FTS lifter which improves the performance of the entire system by considering the following procedures:

• the lifter moves only between requesting and receiving floors

• pools the orders (if possible move a second box as well)

3.4 Design of Experiments

In order to evaluate the improved system the FTS will be tested under normal load. The simulation runs for seven days and the logistics is operating 24 h per day. To test the system the performance of the FTS is checked when the different load scenarios are applied to the logistic area. These are:

• 25% additional volume

• change in the arrival interval time of the trucks (1 h or 7 h) The current arrival schedule for the trucks can be found in Table 3.3.

Arrival Time

3PL 6:00, 10:00, 14:00, 18:00

DC 6:00

Table 3.2: Current arrival schedule for the trucks at Factory One

3.5 Concept of the Model

At the beginning of the thesis project a decision diagram was proposed as the basis for the new control.

In this diagram all possible routes, e.g. from ground floor to first floor, would be mapped in combination with three load scenarios (no box, one box, two boxes). To be more precise the position of the boxes of the lifter, right or left side, would also be taken into account. Even if different routes for the boxes on the lifter are neglected the control would need to describe 24 different scenarios (6 routes x 4 load scenarios). The resulting control would be very complex and the implementation prone to programming errors. Furthermore, the control would not be flexible, as with every change or extension all additional scenarios would need to be specified.

However, a simpler solution to the problem is found by abstraction. The abstracted model is shown in Figure 3.1. Instead of defining every possible route between floors the problem is abstracted to a transport between two different levels. the start floor and the target floor. The direction of travel is derived from these two. If the target floor is above the start floor, the lifter moves up. In the other case the lifter moves down. Initially, the route is determined by a list of all arriving boxes into the system. In this list the start floor and the target floor of each box are recorded. If the lifter is empty the lift will move to the start floor of the first box on the list and load it onto the lifter. The first box will always be loaded according to the FIFO principle. Therefore, in comparison to the old control a box will not wait indefinitely if the control is running into a dead end loop in which the lifter always carries boxes from the same floors. Once the

(38)

first box is loaded, the control checks whether there is a box, which can be loaded onto the other side of the lifter and needs to travel into the same direction. If the second place remains empty and there is an intermediary floor (in this setup only the ground floor) another box can be loaded onto the lifter or the second box can be unloaded if it has already reached its destination. At the target floor the box will be removed. If the second box has not yet reached its target floor the movement will continue to its target floor. Once the FTS lifter is empty the process starts anew.

Due to the abstraction the model can be adapted more easily. For example any number of floors can be added to the model without changing the control. Furthermore, by extending the loading procedure also a varying number of smaller boxes (specification of the lifter: a maximum of four, two per side) could be loaded.

Start floor Intermediary floor Target floor

1) Go to start floor 2) Load

3) If possible load second box 4) Move in the

direction of target floor

1) Check if second box is at its target floor 2) If true, unload 3) If possible load

next box on the list with the same target floor

1) Unload boxes 2) If necessary,

move to the target floor of the second box 3) Find new start

floor Boxes to Transport

1. Box 2. Box 3. Box 4. box

FIFO

Figure 3.1: Concept of the FTS control

3.6 Data Collection

In order to implement the model no additional information needs to be gathered. The necessary information such as the geometry of the system are already implemented by [8, 16].

3.7 Modeling

In the following the implementation into plant simulation will be described. The old model of the FTS is shown in Figure 3.2. The three lifts are represented by conveyors and the lifter moves between the floors are represented by entries and exits on each side. A special case is the temporary storage in the cellar.

Here the material is put into the FTS and taken out of it by a special stations The movement of the lifter is controlled by sensors on each level which triggers the method "Hebersteuerung" (lifter control). The lifter is additionally called to the different floors by reading the list "Rufliste" (calling list). In this list the floor of all arriving boxes is noted when they enter the FTS. In addition, upon entry the "Eingangsliste"

is filled for each box with the following information: unique identifier, entry side, target floor, target side and the direction of travel. This list exits for each floor separately. The various methods are explained in more details in the following sections.

20

(39)

3.7 Modeling

Figure 3.2: Old model of the FTS in plant simulation

3.7.1 Attributes

In the old model the attributes of the entries and exits such as the side or the floor are read from their name. In order to improve the clarity of the code, those information are set by attribute managers in the new model (cf. Figure 3.3). There are two attribute managers one for IN (incoming boxes) and one for OUT (outgoing boxes). "Etage" refers to the floor and "Seite" to the side. The "StichNr" refers to the number of the lift and the position refers to the sensors which are marked with lines on the conveyor and which are closest to the entry.

Figure 3.3: Attribute manager for the entry and exit of the boxes

3.7.2 Lists

In order to simplify the information flow within the FTS each lift only has one list for arriving boxes ("Eingangsliste") regardless of the floor. In this list the following information is collected:

• "Gebinde": unique identifier of the box

(40)

• "Seite": entry side, left or right for ground floor and first floor; for the basement as their is only one side of entry

• "Zieletage": target floor

• "Zielseite": target side

• "Richtung": direction of lifter movement. up = false. down = true

• "Startetage": start floor

• "Position": position of the corresponding sensor (0 = basement, 2 = ground floor, 4 = first floor) The positions are used to derive by subtraction the direction of movement. If the target has a higher number than the start the lifter is told to go up. The mathematical rule is simpler as not every possible route between "UG" (basement), "EG" (ground floor) and "OG1" (first floor) needs to be defined by if-clauses in the methods.

Figure 3.4: List for arriving boxes

3.7.3 Arriving Boxes

The boxes arrive in the system by a line element (e.g. "FoerderstreckeIn 1 EG L") and are controlled upon entrance by the method "Annahme" (acceptance). First the attributes are initialized using the attribute manager described above.

The method distinguishes between the material coming from the logistic area or production from the boxes in the temporary storage. In the first case, the arriving side is read from the attributes. The target floor is set to the sensor position "1" for the basement and the target side is left blank as there is only one exit in the basement and the direction is set to "true" which is the equivalent for down. In the second case, the sensor position of the target and the target side is derived from the global list "Liste Weg ID", which maps each route with a unique identifier and its position. As all boxes need to travel up the direction is set to "false". In both cases, the right "Eingangsliste" for each lift is selected and filled with the required information. Furthermore, the global list for the FTS is updated and a space is reserved for the box. By this reservation the system prevents that too many boxes enter the FTS at the same time. In addition, all available information are written into the list "Auswertung" (evaluation) in order to analyze the system behavior.

The main change is the simplification of reading the attributes. Formerly, the attributes were extracted from the name of an object. In addition, only one list is kept for each lift instead of for each incoming line. Furthermore, the list "Rufliste" (calling list) was abandoned which further simplifies the model.

The implications of the last change are discussed in the following section.

22

(41)

3.7 Modeling

3.7.4 Control

The actual control of the lifer is done by the method "Hebersteuerung". The method is called every time the lifter passes one of the five sensors (indicated by red in Figure 3.13) on the lift. The next target of the lifter is set by either the internal list of the lifter or the list "Eingangsliste". Whenever a box is set onto the lifter it is copied to the internal list together with its target floor and its target side. Based on the first element on the list the direction of movement and the target floor are set. If there is no element on the list the target floor is set by the start floor of the first box on the list "Eingangsliste".

As the control differs significantly in the logic and code from earlier versions the programming is explained in details. After the variables have been declared, the parameters and lists are initialized (cf.

Figure 3.5). Unfortunately, the number of the lift needs to be taken from the name of the object as the method is not triggered by the lift but by the sensor and the attribute cannot be given to the sensor. In this case the right lift and "Eingangsliste" are selected. Furthermore, the start floor of the first element on the

"Eingangsliste" is read and declared as the variable "Anfrage" (request).

Figure 3.5: Parameter setup for the stations in the FTS

From Figure 3.6 it can be derived that the lift is paused if the internal list of the lifter and the "Ein- gangsliste" are both empty. By doing so the lifter only moves when it is necessary.

Figure 3.6: Routine for pausing the lifter

Next the next target ("naechstesZiel") of the lift is selected. Either from the target floor of the first box on the internal list or if the list is empty the start floor from the first element on the "Eingangsliste". The code is shown in Figure 3.7.

Figure 3.7: Selecting the next target

Afterwards the methods decides on the direction of movement based on the next target as can be seen in Figure 3.8. If the next target is above the read identification of the sensor which triggered the method.

the lifter moves upwards. If the lifter is on the first floor on the basement the direction is set in the inspect section to avoid that the lifter overshoots. In addition, for every sensor ID the floor ("Etage". e.g.

(42)

ground floor ("EG")) and position (e.g. ground floor incoming ("EG in")) are defined. Based on these information the lifter can load and unload the boxes correctly.

Figure 3.8: Representation of floors in the control

In the next section of the code (cf. Figure 3.9) the lifter is stopped once its target is reached. Otherwise the lifter would just move in circles without loading and unloading.

Figure 3.9: Halting the lift when next target is reached

In Figure 3.10 the preparation for the loading and unloading procedure are stated. If the current sensor ID exits in the internal list an internal request is filled. The internal request is used later to trigger the unloading. If the "Eingangsliste" has at least one entry an external request is filled. This request is used to trigger the loading procedure.

Figure 3.10: Request for loading and unloading

The unloading procedure is shown in 3.11. If the internal request is true, meaning that the floor is the

24

(43)

3.7 Modeling

same as the target floor in the internal list, the lifter is stopped. In order to read all entries in the internal list, and therefore to unload all eligible boxes, a repeat until loop is entered. If one element is found the exit station is selected depending on the floor and the side. After a time delay to simulate the deboarding the element on the internal list is transferred to the exit station and deleted from the list. After all eligible objects have been unloaded the lifter continues its travel to the next sensor, unless boxes can also be loaded. This procedure is described in Figure 3.12.

Figure 3.11: Unloading procedure

If an external request exists and the current positions matches this request the lifter is stopped. If there is space on the lifter the method for loading the boxes is triggered as shown in 3.12. This method is shortly discussed in the next section. The direction ("Richtung") needs to be set again. Otherwise the lifter would only bounce between the two sensors at each floor. After the boxes have been loaded the lifter continues.

Compared with earlier versions the routines for controlling the movement of the lifter are changed significantly. Instead of checking in the global list whether the lifter needs to move, it only moves if there are boxes in the system. Furthermore, the lifter only travels if it has a target either delivering a box or picking one up. The procedures for the loading and unloading of material have only been improved slightly.

3.7.5 Loading

The called method for loading the boxes into the lifter remains the same as in earlier versions. Its task is to move the box onto the lifter and to determine its exit buffer. Furthermore, it handles the blocking of the two available spaces on the lifter.

(44)

Figure 3.12: Procedure for loading the boxes is triggered

3.7.6 Validation and Documentation

In order to validate the model several lists and graphs have been added to the model. To check on the data of all entering boxes, these are not only recorded in the "Eingangsliste" but also in a statistics list. In this list all available attributes are recorded, e.g. the type of the box, its destination and its identity. Although the last is deleted when the box leaves the system the other parameters can be used for analyses which do not require an identification of each box. In order to monitor the material flow a Sankey diagram is added and can be used to find boxes which go astray. To test the limits of the logistic area a breakdown of the FTS is simulated. This breakdown is recorded by the list failures of the FTS. In the list the start and the end of the breakdown is shown in order to evaluate whether the settings of the system are correct.

The performance of the three lifts is analyzed by a utilization graph. Especially the working share of utilization is important to detect the limits of the system and to compare it to real life.

Validation tool Purpose

Statistics list record information of all boxes entering the system to analyze their parameters, e.g. validating the distance traveled Sankey diagram check material flow between elements

List for failures of FTS check the mean time between failures (MTBF) and the mean time to repair (MTTR) Resource statistics analyze utilization of the lifts

Table 3.3: Current arrival schedule for the trucks at Factory One

3.7.7 Summary

The new model is shown in Figure 3.13. As the most changes affected the control method the only visible changes are the reduction of lists for incoming boxes and the additional tools for validation.

3.8 Validation

In the following, the implementation of the model in plant simulation is checked for consistency and the model is validated. As the most significant changes affect the control its behavior is discussed in

26

(45)

3.8 Validation

Figure 3.13: New model of the FTS in plant simulation

details. Multiple runs are unnecessary as there are no stochastic elements in the model as failures and other random elements are not implemented.

First of all, the overall performance of the FTS is checked. As can be seen in Table 3.5 the new control transports the same amount of boxes as the old control. The difference is very small and is caused by the higher performance of the new control which can transport a few more boxes at the end of the simulation.

As both systems operate in a steady state this difference can be considered as insignificant and the model as valid in terms of the overall output.

moved items lifer 1 lifter 2 lifter 3 total

old control 34’095 15’182 11’112 60’389

new control 34’146 15’185 11’114 60’445

difference between old and new controls 0.15 % 0.02 % 0.02 % 0.09 %

Table 3.4: Moved items by old and new control

Based on the model all boxes are first moved to the basement and to the temporary storage before being transported to their destination. This holds true for the model as can be derived from 3.5. It shows how many boxes are transported from which floor (left column) to which floor (first line) across all three lifts.

As material is transported only between the floors and the basement the logic is implemented correctly.

The small difference, that fewer elements leave the basement than enter it, is due to a small stock of items which are in the temporary storage at the end of the simulation. Therefore the difference is not significant and the model is valid concerning the material sums.

Based on the FTS statistic list as shown in Table 3.5 the minimum distance traveled for each lifter was calculated. As the start and target floor of each box is recorded in this list the distance between these two can be calculated. It can be considered as the minimum traveled distance of the lifter which does not take the traveled routes without a load to the start floor into account. In order to compare the simulated distance with the calculated it can be assumed that the difference is smaller than factor 3. This factor accounts for the empty routes which could be longer than the routes with load (e.g. empty route first

Modeling and Optimization of the Production and Logistic Concept of the Siemens "Factory One" in Zug