Quay cranes on modern container terminals have the possibility to lift two 40ft. containers side by side. In order to be able to perform this type of lift, the crane must be equipped with a tandem spreader. As this type of spreader cannot be used for performing certain lifts, such as the handling of hatch covers, switches between single spreader and tandem spreader are inevitable. However, there is no method in current literature nor in practice to determine when the spreader changes should be performed. Therefore, the aim of this research is to develop a model that calculates the optimal moments to switch spreaders. As the performance of tandem lifting depends on the cycle times, a cycle time prediction model is developed first. This Artificial Neural Network model predicts the cycle time based on the type of lift, the actual position of the container(s) on the ship, the weight of the load and the current wind speed. The predictions yield a Mean Absolute Percentage Error of less than 11%. Afterwards, the cycle time prediction model is used to develop an optimal spreader switching strategy model in the form of a Mixed Integer Linear Programming model. A case study, in which several different bay layouts of a container ship need to be handled, shows a decrease in handling time of up to 22% compared to the traditional spreader switching strategy.

For the shared mobility service providers, it is essential to have an economic operation and therefore, it is necessary to maximise their fleet utilisation. Of influence on the latter is the availability of the spare parts. The topics for which spare part forecasting research is already performed are different in several ways compared to the field of free-floating shared vehicles. The many different users with each different driving skill, wear and tear by operation, events as vandalism and traffic accidents and constantly standing out on the street result in sudden failures of parts. Three different forecast models are presented in this paper. The effect of including demand generating factors in the forecast approach is analysed. The methods show that a better forecast performance can be achieved while including demand generating factors into the analysis. The forecast accuracy can be further improved and some suggestions for improvement are given in the paper.

This research investigates ways to maximize the outbound handling and freight to customer operations efficiency based on the real data of The Kraft Heinz Company. A cost to serve model has been developed to evaluate the performance of various design alternatives deriving from the field of "quantity adjustment" as the dependence of cost efficiency upon labour intensity at the warehouses and the truck volume utilization indicated. The research concludes upon the selection of the optimal design alternative to be implemented per customer as well as upon which design alternative is more suitable for clusters of customers that share specific characteristics.

Heineken's goal with respect to safety is to reduce the number of accidents to zero. Such a drastic reduction can only be achieved by considering safety as early as in the design phase, when most impact can be made. Therefore, the purpose of this research is to enable the easy generation of different design concepts and evaluate them with regard to safety. Since most accidents currently occur in Heineken's warehouse loading areas, this is the area considered in this thesis. In order to evaluate the generated designs with respect to safety, safety has been quantified into a single value, the risk level. This risk level is the result of the possible consequence of an accident (the impact), the number of times an activity is expected to occur (the frequency) and the chance that this activity results in an accident (the probability). These values depend on the type of accident, or hazard, which is any unsafe condition or potential source of an undesirable event with potential for harm or damage. The hazards corresponding to the top three most occurring accidents have been identified and have been evaluated. For the generation and evaluation of design concepts, a parametric model has been developed. By changing parameters, designs with different operating processes, equipment types, layouts, and more, can quickly be generated and visualised for the designer. Also the movements that occur in the warehouse area can be simulated. The model counts the number of times a hazard occurs and determines the corresponding frequency score. The scores for the impact and probability are based on expert opinion. The risk level is subsequently calculated. Using the parametric model, the risk level resulting from the twelve most occurring hazards in the warehouse loading area of Heineken is evaluated for different scenarios. The findings show that the current method that is used to determine the frequency score is not suitable for activities that typically occur a lot, which is the case in the warehouse loading area. Therefore, an alternative method is proposed, which uses an exponential distribution to score the number of times that a hazard can occur. By analysing design choices regarding the type of picking process (directly from warehouse, using cross-docking or using pre-staging), the level of automation, the number of aisles, the distance between loading points and designing crossings on a different level, the preferred design choices are found.

In the competitive offshore wind installation market, contractors like DEME strive to optimize operations. This thesis centers on the optimization of the support vessel fleet composition, which consist of a num- ber of walk to work vessels, and are a crucial part of inter-array cable installation. By fine-tuning this composition, operational expenses can be reduced by 10-20% to current industry practices. The cable installation process involves a complex set of operations, each necessitating a crew to be present on the foundations. Support vessels play a central role in routing of these crews and their equip- ment to these foundations.This thesis introduces an innovative approach that integrates operational scheduling and crew routing into a single formulation for optimizing the fleet of walk-to-work vessels. This hybrid model combines elements of a continuous-time rich multi-visit multi-period Vehicle Routing Problem with a time-varying Resource Constrained Project Scheduling Problem. It also factors in the substantial impact of weather conditions on offshore operations by accounting for variable weather win- dows in each scheduling period. The formulated model is rigorously verified and validated to ensure it closely mirrors real-world sup- port vessel behavior. Although it slightly underestimates fuel consumption, this discrepancy is deemed acceptable given its minor role in the overall objective. Sensitivity analyses highlight the critical impor- tance of accurate performance data for the cable laying vessel, which significantly influences the model’s outcomes. However, the model is found to be most effective for modeling a single cable string compris- ing 6-8 turbines, with scalability issues arising when attempting to expand beyond this scope. A focused case study delves into the impact of various weather conditions and inter-array distances on the optimal vessel composition. The study evaluates two types of walk-to-work vessels, individually and in combination. Results reveal that, under the assumption of zero downtime, the industry norm of chartering cheaper vessels is cost-effective, while the pricier vessel results in a 10% costlier solu- tion. As weather conditions worsen, a composition of costlier vessels proves more cost-effective over the scheduling horizon. Such conditions are to be expected in far offshore locations, especially on the cheaper vessels, which have lower workability limits. The study identifies potential cost reductions of up to 25%, with even marginal downtime conditions yielding 10-20% reductions to the industry standard. Moreover, delays imposed on the cable laying vessel are significantly reduced when utilizing a compo- sition that includes at least as one of the more expensive vessels. In summary, this thesis establishes a foundation for optimizing support vessels in offshore wind installa- tion. The presented model introduces a novel framework, combining multi-visit routing with time-varying resource scheduling, while considering shared vehicles and coupled routing and scheduling over a multi- period horizon. For regions prone to harsh weather conditions, such as those further offshore and during winter months, it is advised to utilize more expensive vessels that have higher workability limits and bet- ter performance figures. Additionally, there appears to be limited justification for simultaneous use of multiple vessels during the cable installation, as the added costs do not seem to outweigh the marginal improvements in installation duration. Future research should focus on refining solution methods for the proposed formulation and incorporating crew transfer vessels into the fleet composition.

The performance of a product is highly influenced by the users of the product. A Combined Performance Measurement Model of a Shared Rental Vehicle is created to analyse the influence of a user on the vehicle performance. A Performance Measurement model is created to analyse the performance of a Shared Rental Vehicle. User Profiling is used to create four user profile perspectives. These models are combined in a Combined Performance Measurement Model to analyse the relation. This Performance Measurement Model propose a shift in user population to gain optimal or target performance of the Shared Rental Vehicle.

Dynamic pricing can be used for better fleet distribution in free-floating vehicle sharing (FFVS), and thus increase utilization and revenue for the provider by reducing the supply-demand asymmetry. Supply-demand asymmetry refers to the existence of an undersupply of vehicles at some locations at the same time as underutilization of vehicles at other locations. We propose to use dynamic pricing as an instrument to incentivize users to rebalance these vehicles from low demand locations to high demand locations. Despite significant research in rebalancing vehicle sharing, literature so far lacks experimental results on dynamic pricing in free-floating vehicle sharing. We propose to use an algorithm that minimizes the differences in the idle time of vehicles. The algorithm is tested in a real-life experiment that was conducted in cooperation with an FFVS provider. The experiment shows that even slight dierences in pricing and a novice algorithm can already influence user-behavior to counter the supply-demand asymmetry. Improving the existing algorithm by doing more experimental research is advised to further uncover the potential of this strategy.

With the increasing focus on project planning in engineer-to-order(ETO) companies, tracking the progress of the design phase of a project is essential for effective planning. Such progress tracking currently is done through the regular feedback of discipline leads, which are intrinsically subjective. The information thus collected is then plotted in the form of so-called S-curves, to represent the progress over the course of the project. The present method tracks progress of the manufacturing engineering phase based on how much mass is designed with respect to the design weight of the equipment. This so-called mass curves are used and fitted to an S-curve and represent the tracked progress. It is shown that the fitted mass curves represent the progress of the engineering phase with a systematic average error of 25 days. Additionally, mass data retrieved from completed projects are used to predict the S-curves shape of the ongoing project. The weighted average of tracked and Predicted S-curves is then used in a Monte Carlo simulation in order to estimate the progress of the ongoing project. The method was found to have a tracking accuracy of 0:6 4%, which translates into an uncertainty on the project duration and therefore on the equipment delivery date, by 2 weeks. This uncertainty is found to be the upper bound of typical grace periods agreed with most clients. The present results might offer a new outlook on progress tracking on the progress tracking of manufacturing engineering activities. Further research shall be conducted to address the progress of earlier engineering phases equally.

The present study was designed to gain a deeper understanding of RTI control and Supply Chain Coordination, with a specific focus on the practical challenges experienced in the reverse logistics by asset-owner and asset-users. A case study at EPS - an RTI owner - is performed and a practical problem has been addressed with a design science and TIL systems engineering approach. It was observed that a huge amount of deposit is allocated to a retailer's generic account. The current state analysis found that this problem is mainly due to inefficiency in the current identification process with physical SSCC labels. The design goal of minimising the "Delta" -the number of unidentifiable return packages- was defined. Ways to redesign the reverse logistics were determined with the aim that all return packages can be allocated to the retailer at shop level. Based on the future state requirements and the design goal, the concept of matchmaking is developed as a way to achieve the design goal. This thesis shows that a matchmaking system can improve the traceability of return packages and that it has the potential in bringing the Delta of return packages with missing shoplink to zero. The matchmaking system is defined as the framework that ensures matchmaking, which uses a "key" generated by the sender to represent the return package. If the receiver can find the key upon arrival of the return package at the depot, the sender can be identified. Seven matchmaking systems were considered. Five alternatives use unique tray identities as key. One alternative uses the load carrier as key and the last alternative uses the return package identity as key. The validation results show that an "all read" or "reading of all individual trays" is not a requisite for a working matchmaking system. By contrast, as long as a certain ratio of reading at two locations is reached, an all-read scenario can be mimicked. The assessment investigated the instances in which zero mismatch take place. Results show that the higher the data capture capability of both the sender and receiver, the higher the chance a match can take place and the smaller the chance of a mismatch. This thesis creates insights on requirements for enhancing traceability of RTI's in the return chain and developed a matchmaking concept that can address the practical problem of returns without traceability of shop origin. The developed matchmaking concept is the outcome of an analysis of the current state and makes use of data elements that are already being collected in the database, in the case of EPS. The study addresses how collected data can be leveraged for enhanced RTI management in the reverse logistics and may inspire practitioners to face challenges with a similar lean approach.

Jumbo Maritime is a company specialized in handling and transporting substantial cargoes which did not fit into a container. To compete in the heavy lift shipping market, a well-balanced fleet composition is essential. In the current situation, the heavy lift cargoes are transported by a fleet consisting of ten vessels, having a varying crane lifting capacity from 3000*10^3 to 650 *10^3 kilogram (Class K, H, J and E). However, in the past five years, it has become apparent that the deployment of the vessels based on the heaviest item (between 75-100% of the capacity) has only been needed in a meagre five percent of all transportations. The installed crane capacity on the Jumbo vessels are an overcapacity compared to the demand of the market. During this research, a decision support tool is developed to improve the match between the demand from the market in the heavy lift cargo industry and the supply of the heavy lift shipping vessels. This will result in an improved fleet composition that maximizes profit. Subsequently, the tool has been examined to see whether it can henceforth be applied to the fleet composition to support Jumbo Maritime in its leading role of supplying heavy loads.The Maritime Fleet Size & Mix Problem model was used to find the optimal fleet composition. The goal of this model is to maximize profit, achieved from transport orders based on reduced cost of the vessels. Considering that the proceeds of a load depend on market demand, the transport data of the past five years has been used as an input parameter. A fleet is chosen by taking into consideration both the technical and financial aspects. The technical aspect is the maximum combined crane capacity installed on the vessels. The financial aspects of the vessels exist of capital costs, labour costs, fuel costs and maintenance costs. The result of this model will lead to an optimization in the distribution of the vessel.Subsequently, the vessels that are at the end of their life will be replaced in the near future. The Vessels classified under H and E will be replaced by a variety of the newer J-light classes. The decision about the composition of the new vessels in the Jumbo fleet should be supported by the optimization model. The new vessels will be used as input in the model, together with the prediction data of possible future market needs. These future predictions will consist of three scenarios: 1) the demand will stay the same as it has been for the past five years; 2) Demands will rise with 6% due to an increase in the oil and gas industry; 3) a 35% increase in the renewable energy related market. The result of the three scenarios will be a new fleet composition with a corresponding profit.An optimal fleet composition will be the result when the transport data of the past five years, together with the current fleet characteristics, are used as input for the Maritime Fleet Size & Mix Problem. The current fleet composition [2:4:2:2] is in the optimized situation as followed [1:2:1:8], whereby the theoretical gain increases from 108 million to 181 million. When the future scenarios one, two and three together with the new vessel types (K: J: J-900: J-800: J-700: J-600) are introduced, new optimum fleet compositions arise. The new optimized compositions are as follows: scenario one: [1:1:0:1:0:6] with a profit of 56 million, scenario two: [1:1:0:1:0:6] with a profit of 68 million and scenario three [1:1:0:1:0:6] with a profit of 58 million.The results of the Maritime Fleet Size & Mix Problem show that the profit can be optimized with an alternative fleet composition. The mathematical model determines the composition of the fleet by using the transport data from the past five years. The optimization model is subjected to a number of assumptions to approach the reality. By also using the Maritime Fleet Size & Mix Problem model with the introduction of the J-light and combining it with the three different future scenarios, the vessel type with a combined crane capacity of 600 tonnes appears to be the most efficient in the J-light class.

This report describes a research performed to investigate in which way it is feasible to dynamically alter the floor plan and desired capacity of airport baggage handling systems by making use of autonomous individual transport robots. By researching the state of the art of both conventional baggage handling systems and transport robot systems, a new baggage handling solution called the Baggage Robot Concept is proposed. A simulation model of this Baggage Robot Concept is developed to evaluate the performance of this concept in the sorting process of a baggage handling system at medium-sized regional airports. The research identified that the most important design elements for the Baggage Robot Concept are the number of robots and the floor layout configuration. The Baggage Robot Concept turns out to be a feasible concept. An important design choice that is identified in the research is the trade-off between manoeuvre space of robots and the location of charging and storage positions. This study has been a first exploration in integrating autonomous robot systems in baggage handling systems, contributing to future proof and cost efficient operations.

This thesis has provided insight into how machine learning can be beneficial to path planning in container terminals. Path planning algorithms can be used in environments with automated vehicles. A well known algorithm is the A* path planning algorithm, which is the fastest optimal path planning algorithm under satisfied conditions. However, the behaviour of a container terminal is unknown beforehand, costs can change over iterations. Therefore, Liu et al. [Liu et al., 2019] and Keselman et al. [Keselman et al., 2018] show the advantage of combining A* with Machine Learning. This way, the exploring part of the ML algorithm is combined with the fast andmore precise properties of the A* PP algorithm. This thesis has proposed the machine learning algorithm Vehicle Aware Reinforcement Learning Path Planning Algorithm VARLPPA. This algorithm uses Monte Carlo Control method. This is a model free approach, which has been shown in both experiments to find more efficient solutions in exceptional situations.

Formation control of autonomous surface vessels (ASVs) is researched extensively in the last years, since it has promising applications. However very few designed strategies have been validated during real experiments. In this research, two control methods for distributed leader-follower formation control are proposed: A Nonlinear Model Predictive Control (NMPC) method and an MPC method using Feedback Linearization (FL). These two designed methods are compared with a conventional Proportional-Integral (PI) control method. The performance of the proposed strategies is evaluated through simulations and real experiments using small-scale Roboat units. Both designed control methods outperform the conventional PI control method in simulation and real experiments.

This paper deals with the food flow for intercontinental KLM flights. The aim of this research is to design a framework to optimize the processes of food assembly and subsequently apply the framework to increase the throughput time at the intercontinental department of KLM Catering Services. A discrete event simulation model is created to compare key performance indicators for different design alternatives. In this simulation model different amount of equipment is handled together with different settings for a robot. With the use of this simulation model a combination of design alternatives is proposed.

Supply chains are to be found in almost every corporation around the world. When tryingto manage a supply chain often two types of disturbance can jeopardize the effectiveness of a supply chain. These are supply and demand uncertainty. This thesis tries to investigate the effects of those two types of disturbances in a supply chain and seeks to find a remedy. When taking a closer look at supply chains, robust supply chains are considered to withstand a higher level of supply and demand uncertainty. Robustness knows many definitions but is mostly characterized by the responsiveness and adaptability of a supply chain. Responsiveness implies a degree of quick reaction to sudden changes, while adaptability is defined as the capacity to deal with new circumstances. This thesis tries to relate robustness according to a level of resistance of supply and demand uncertainties. The supply chain of the royal dutch land forces is prone to such sudden changes. Therefore this research incorporates the context of a military supply chain to perform experiments on. The military supply chain in operational areas is selected and identified as a 3-echelon-multiple tier-supply chain, where the implementation of a networked supply chain can result in an increase in robustness. The research uses three key performance indicators to ”measure” a degree of robustness. With the use of a multi-agent discrete-event model that incorporates a highly dynamic environment including real-time transportation and order handling simulations. The model also contains an nteger supply optimization module using CPLEX and a distance-based swarming algorithm to emulate the lifelike situation of the supply chain concerning order handling and ”Notice to move”. Three experimental setus have been created that try to quantify robustness levels according to networked supply chain operations while coping with demand and supply uncertainty. First, a sensitivity analysis is performed, revealing the basic relationships between the indicator and emand/supply fluctuation. Second, an extensive model testing according to three different load cases in uncertainty is performed. Thirdly a case study specified for a military context is done. It is found that robustness can be increased by increasing the amount of networked forward supply nodes and interesting dependencies are revealed between the amount of forward supply centers and robustness. The findings of this research can be applied to other industries that have a similar supply chain design.

This research focuses on identifying and evaluating ways to improve the sustainability of solid waste management (SWM) in semi-public spaces using the Efteling theme park as a case study. To the best of the author’s knowledge, this study is the first to apply an integrated waste management approach to optimize SWM in this context. The approach consists of assessing the integrated environmental and economic burdens of the collection, transportation, sorting and treatment processes of a SWMstrategy. Approximately 10 alternative high potential SWM strategies were composed and selected out of a very large pool of possible SWM strategies. The environmental and economic impact of these high potential SWM strategies for the Efteling case was subsequently assessed using the WARM LCA model and a custom economic assessment model. The results reveal that emission reductions of up to 190 ton CO2 equivalents (TCO2E’s) per year, relative to the current strategy, can already be achieved by separating and recycling more fractions from behind-the-scenes (BTS) waste. This corresponds to an increase in yearly avoided SWM system emissions of about 25% for the Efteling case (from a life-cycle perspective). When public/semi-public (PSP) waste is also included, emission reductions of up to 800 TCO2E’s per year can be achieved for single-fraction separation (e.g. PMD or PET) and up to 960 TCO2E’s per year for two-fraction separation (e.g. PMD + paper or PMD + cups). This corresponds to major increases in avoided SWM system emissions of 110% and 131% respectively. It was also found that small interventions in the transport and/or treatment waste management components can make a big difference in the environmental and/or economic impact of a SWM strategy. A majority of the alternative SWM strategies has an eco-efficiency (emission reduction cost-effectiveness) ranging from €39 to about €140 per TCO2E saving. This eco-efficiency is (much) higher than that of a range of benchmarks such as the eco-efficiency of solar-pv panels at a non-industrial scale, wind turbines at sea and the eco-efficiency of office building insulation. This indicates that optimizing waste management should be given more priority in (scientific) research as well as in practice.

In this report, supply chains of factory outlet stores are analyzed. Here, the concept of factory outlet stores have been investigated in literature to be able to understand which goal these stores serves. Based on a pull based supply chain perspective, replenishment has been investigated. Here, demand forecast, inventory, order ful_llment and transport have been considered. Here, transport models from literature are introduced to see which factors are considered and for which purpose these models can be used. Also, the performance of supply chain has been discussed, which consists of _nancial as non _nancial aspects. A case study will be done at the factory outlet store of L'Oreal. Here, the supply chain has been analyzed, as the catalogue and type of products they o_er. The replenishment policy has been analyzed, in which the transport process from the CPD warehouse to the store will be investigated. Here, a model will be designed to be able to schedule the deliveries to the store in a more cost e_cient way. An analytical modelling method is selected to develop a model which captures the mentioned transport process. Here, aspects of models in literature are evaluated to be used in the transport model for scheduling deliveries. The model is developed as an integer mixed linear programming model and will schedule the amount of pallets per delivery to a storage location, in which two storage location are presented. This model is described as the delivery scheduling transport model (DSTM) and is checked on implementation in MATLAB in a veri_cation step. The DSTM has been con_gured for the case study at L'Oreal, which is checked by a validation step to see if the model meets the system requirements. Then, experiments with di_erent inputs are executed to see if savings can be realized by scheduling the transport via this model. This is realized in all the inputs scenarios. To see if more savings can be realized, sensitivity analysis have been performed to investigate how parameters can be tuned to realize lower cost. A conclusion will _nalize this research to conclude that replenishment orders can be scheduled in a more cost e_cient way by the DSTM, which is the result of a case study of L'Oreal.

In the construction industry, it is common to predict manufacturing times of structural elements based on the experience of shop managers. With the increasing use of Building Information Models (BIM) and data analysis in manufacturing processes an opportunity for improving the accuracy of predicted manufacturing times arises. This research proposes a general prediction model based on physical properties of structural elements extracted from BIM, along with manufacturing times of manufactured elements. The proposed Support Vector Regression Model Tree (SVRMT) is tested in both hypothetical and real case scenarios. Through validation in real case scenarios, the SVRMT prediction model turned out to be a significant improvement in terms of prediction accuracy, compared to the current experience based approach.

Abstract Purpose - The aim of this paper is to propose a method for performance measurement of the livestock feed industry from an environmental and economic perspective. There is a knowledge gap both in the literature and society in the field of performance measurement of the livestock feed industry with environmental concerns. Livestock feed companies need support in the decision-making process to produce environmentally sustainable livestock feed at as low as possible costs. Design/methodology/approach – An environmental performance index for livestock feed is constructed based on techniques of the min-max transformation, the analytic hierarchy process and simple additive weighting. The verified environmental performance index for livestock feed is brought into practice; different scenarios, which all represent a feed composition, are quantitatively compared to a benchmark feed composition. The environmental performance index is validated with an uncertainty analysis and sensitivity analysis. Findings – The constructed environmental performance index for livestock feed is assessed by comparing nine scenarios against a benchmark feed composition. The results indicate that the environmental performance index for the livestock feed industry is technically feasible and effective to measure the performance of the livestock feed industry from an environmental and economic perspective. Research implications – The constructed environmental performance index for livestock feed can serve as a decision-making tool for livestock feed companies. As a response to climate regulations and a market pull, this paper provides a tool for livestock feed companies to select livestock feed compositions with the best performance based on environmental sustainability and economic performance. Originality/value – A new performance measurement method is designed for an environmental performance index for the livestock feed industry. From the literature, environmental sustainability variables are identified. The data on the environmental performance are obtained from a public database as well as a livestock feed company. The created index can contribute to decision-making in the livestock feed industry.

Globalization is heaving its impact on many field. For the field of warehousing this development, in combination with the actual shortage of labor, means that many warehouse owners are forced to reduce their warehouse costs in order to stay competitive with warehouses in low-wage countries. On the other side the warehouse owners are pulled by recent technical innovations, which have reduced the drawbacks of automation. One of these recent innovations, is the mobile fulfillment unit-to-unit order picking system, which is presented in this paper. To make this system more affordable and hence to increase the competitiveness of warehouses in western Europe, research is required optimizing each of the system parts. This research focuses on the order processing part of the system, which has changed considerably compared to existing mobile fulfillment order picking systems. In this system units are brought to two sides of a stationary order picker, who then transfers products from one unit to another unit. The order processing problem covers the sequencing of the unit visits at the order picking station, the batching of these units and the determination of the products that should be transferred. Using insights from literature, several approaches are proposed to each of the problem parts. Each of these approaches is then tested and validated. Combining the approaches for each problem part leads to solutions addressing the full problem. All these solutions are tested against each other using a data set of one day of an existing warehouse. The tests are executed for multiple different order picking station configuration and for multiple data set sizes. This leads to an overview of the effectiveness of each of these solutions in terms of computational time and the reduced number of AGV visits. From the results several conclusions can be drawn concerning the optimal solution for a specific warehouse. For an example warehouse in the Netherlands, a significant reduction of 25% can be obtained, regarding the number of visits, compared to the baseline approach. Which means the required number of AGVs can be lower, the investment costs for this system can be reduced and hence the competitiveness for this warehouse can be increased by implementing one of solutions presented in this paper.