Potential improvements in a container terminal through information sharing Master’s Thesis in the Master’s Programme Supply Chain Management ARVID EDFORSS JESPER HANSSON Department of Technology Management and Economics Division of Service Management and Logistics CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2019 Report No. E2019:023 Master’s thesis E2019:023 Potential improvements in a container terminal through information sharing ARVID EDFORSS JESPER HANSSON Tutor, Chalmers: Stefan Jacobsson Department of Technology Management and Economics Division of Service Management and Logistics Chalmers University of Technology Gothenburg, Sweden 2019 Potential improvements in a container terminal through information sharing ARVID EDFORSS JESPER HANSSON © ARVID EDFORSS JESPER HANSSON, 2019. Supervisor: Stefan Jacobsson, Service Management and Logistics Examiner: Per-Olof Arnäs, Service Management and Logistics Master’s Thesis E2019:023 Department of Technology Management and Economics Division of Service Management and Logistics Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: STS cranes together with straddle carriers and a container ship. Printed by Chalmers Repro Service Gothenburg, Sweden 2019 iv Potential improvements in container handling through information sharing ARVID EDFORSS JESPER HANSSON Department of Technology Management and Economics Chalmers University of Technology Abstract Around 90% of world trade is currently transported through the shipping indus- try. The competitiveness is high and companies in the industry are continuously searching for new ways to increase performance and cut costs. Seaport terminals are important actors in the shipping industry since they are the connection between sea and land transport. One problem today at some seaports, which is reflected in literature and at a case terminal, is the limited information that is shared with the manual straddle carriers. This master’s thesis aims to investigate whether increased information in one con- tainer terminal can improve its operational effects. Interviews, observations and tests were performed at a case terminal to evaluate the impact of increased informa- tion sharing for straddle carriers moving containers to and from the ship-to-shore cranes. During the tests, additional information was shared to the straddle carrier drivers through a text message function, with a maximum limit of 38 characters. The operational effect of the STS-crane was measured during the test and the result was evaluated with the usage of a statistical z-test. The findings were, by increasing the shared information to straddle carriers, the op- erational effects were statistically significant that the ship-to-shore cranes increased their performance. If the possibility to share information would increase further than the current limit of 38 characters, the performance of STS-cranes and straddle carriers could possibly be improved even more. Therefore, suggestions for additional information that did not fit in the text message are also presented in this thesis. Keywords: Information Sharing, Container Terminal, Seaport, Straddle Carrier, Ship-to-shore Crane. v Acknowledgements This master’s thesis has been carried out during the spring of 2019 at the division of Service Management and Logistics. The thesis was the final degree project in the master program of Supply Chain Management at Chalmers University of Technol- ogy. The thesis has been conducted in collaboration with one case terminal. Firstly, we would like to thank the case terminal for letting us perform our mea- surements. Our contact person at the terminal for providing us with access and connections. The workers at the case terminal who participated during the inter- views and allowed us to accompany them in their machines during their daily work. We would also like to thank the dispatchers and planners who helped us during our measurements by lending computers, teaching us how the system work and gave us free access to anything we requested during the progression. Finally, we would like to thank our supervisor, Stefan Jacobsson, for guiding and supporting us with feedback. We are grateful for all help he provided and his knowledge contribution within the area whenever needed. Gothenburg, June 2019 Arvid Edforss Jesper Hansson vii Contents List of Figures xiii List of Tables xv List of Abbreviations xvii 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Case terminal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.4 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Review of relevant literature 5 2.1 Intermodal container transports . . . . . . . . . . . . . . . . . . . . . 5 2.1.1 Containers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1.2 Container terminal . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1.3 Operations in container terminals . . . . . . . . . . . . . . . . 6 2.1.4 Equipment in container terminals . . . . . . . . . . . . . . . . 6 2.1.5 Vessel structure . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Yard management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3 Information sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.4 ARA-model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4.1 Actors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4.2 Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4.3 Activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.5 Lean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.5.1 Lean in container terminals . . . . . . . . . . . . . . . . . . . 19 2.5.2 Measure activity time . . . . . . . . . . . . . . . . . . . . . . 21 2.6 Statistical testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.6.1 Boxplot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.6.2 Paired z-test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3 Methodology 25 3.1 Research approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.2 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.3 Research process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.4 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 ix Contents 3.4.1 Observations and interviews . . . . . . . . . . . . . . . . . . . 27 3.4.2 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.5 Statistical test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.5.1 Example of the calculation procedure . . . . . . . . . . . . . . 33 3.6 Reliability and validity . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.7 Earlier experience from seaports . . . . . . . . . . . . . . . . . . . . . 35 4 Results 37 4.1 Mapping of terminal operations . . . . . . . . . . . . . . . . . . . . . 37 4.1.1 Actors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.1.2 Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.1.3 Activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.2 Interviews - Operators on information sharing . . . . . . . . . . . . . 57 4.2.1 Sequence order . . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.2.2 Visibility when loading and unloading large vessels . . . . . . 57 4.2.3 Weight information unloading twins . . . . . . . . . . . . . . . 58 4.2.4 Turn back to previous step . . . . . . . . . . . . . . . . . . . . 58 4.2.5 Automatic sequence changes loading empty containers . . . . 58 4.2.6 Break changes and dual . . . . . . . . . . . . . . . . . . . . . 59 4.3 Information sharing - Test messages . . . . . . . . . . . . . . . . . . . 59 4.3.1 Evaluating information . . . . . . . . . . . . . . . . . . . . . . 59 4.3.2 Information configuration . . . . . . . . . . . . . . . . . . . . 60 4.4 Operational effects of information sharing . . . . . . . . . . . . . . . 60 4.4.1 First test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.4.2 Second test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5 Discussion 67 5.1 Terminal operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 5.1.1 Dispatching and balancing operations . . . . . . . . . . . . . . 67 5.1.2 Reduced dispatcher interaction . . . . . . . . . . . . . . . . . 68 5.1.3 Continuous improvements and employee interaction . . . . . . 68 5.1.4 Tally workers communication and education . . . . . . . . . . 69 5.1.5 Learning organization through lean . . . . . . . . . . . . . . . 70 5.1.6 Call of Vessels and associated network . . . . . . . . . . . . . 70 5.2 Evaluation of shared information . . . . . . . . . . . . . . . . . . . . 71 5.2.1 Additional screen in the straddle carrier . . . . . . . . . . . . 72 5.2.2 Sequence order . . . . . . . . . . . . . . . . . . . . . . . . . . 72 5.2.3 Improvement for the visibility issue . . . . . . . . . . . . . . . 72 5.2.4 Different departments . . . . . . . . . . . . . . . . . . . . . . 73 5.3 Operational effects of sharing information at the case terminal . . . . 74 5.3.1 Test design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 5.3.2 Straddle carrier driving distances . . . . . . . . . . . . . . . . 75 5.3.3 Driving distances and affected crane cycles . . . . . . . . . . . 75 5.3.4 Deviation of the driving distances . . . . . . . . . . . . . . . . 75 5.3.5 Significant difference when sharing information . . . . . . . . . 76 5.3.6 Test interpretation regarding significance . . . . . . . . . . . . 76 5.3.7 Improvement by information sharing . . . . . . . . . . . . . . 77 x Contents 6 Conclusion 79 References 83 Appendix A Interview guide I Appendix B Example of data from StarDriver III Appendix C Code used for boxplot and z-value calculations VII xi Contents xii List of Figures 1.1 Side view of a container terminal adapted by Thoresen (2014). . . . . 2 2.1 Blocked container vessel abstraction (Wilson & Roach, 2000). . . . . 11 2.2 Stowage arrangement for a container vessel, adapted, by Wilson and Roach (2000) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.3 Note: actual storage capacity depends on operational aspects such as required selectivity etc. (Kalmar, 2007). . . . . . . . . . . . . . . . . 13 2.4 Asian (a) and European (b) storage yard layout (Carlo, Vis, & Rood- bergen, 2014). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.5 Storage yard layout for straddle carrier (Carlo et al., 2014). . . . . . . 15 2.6 The ARA-model with its three layers (Håkansson & Snehota, 1995) . 17 2.7 4P model (Liker, 2004). . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.8 Framework which can be used when implementing lean at container terminals (Olesen, Powell, Hvolby, & Fraser, 2015). . . . . . . . . . . 20 2.9 Example over a boxplot (Galarnyk, 2018). . . . . . . . . . . . . . . . 23 3.1 Deductive, inductive and abductive approach (Spens & Kovács, 2006) 25 3.2 A view of the StarDrvier interface used during the measurements. . . 30 3.3 Boxplot over the example data from appendix B. . . . . . . . . . . . 34 4.1 Illustrated sequencing options in container storage rows . . . . . . . . 49 4.2 Boxplot over the first test with and without information. . . . . . . . 61 4.3 Boxplot over the first test with and without information excluding extreme outliers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.4 Boxplot over the second test between the first and second half hour. . 64 xiii List of Figures xiv List of Tables 3.1 All observations during the project. . . . . . . . . . . . . . . . . . . . 28 3.2 All interviews during the project. . . . . . . . . . . . . . . . . . . . . 28 3.3 Table of the two different tests which were performed during the project. 29 3.4 Data origin of the measured KPI’s in the thesis . . . . . . . . . . . . 32 3.5 Results from the example provided in appendix B. . . . . . . . . . . . 34 4.1 The total time that the crane was disturbed by straddle carriers dur- ing the first test. Note that the data have been modified. . . . . . . . 63 4.2 The average distance travelled for each straddle carrier during the first test together with the standard deviation. Note that the distance has been modified. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 4.3 Values for the difference in loading cycle time during the first test. . . 63 4.4 The total time that the crane was disturbed by straddle carriers dur- ing the second test. Note that the numbers have been modified. . . . 65 4.5 The average distance traveled for each straddle carrier during the sec- ond test together with the standard deviation. Note that the distance has been modified. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.6 Calculated values for the difference in loading cycle time in the second test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 5.1 Table of the sample means from the two tests. . . . . . . . . . . . . . 77 xv List of Tables xvi List of Abbreviations AGV Automated guided vehicle FEU Forty-foot equivalent unit KPI Key performance indicator RMG crane Rail mounted gantry crane RTG crane Rubber tyre gantry crane SC Straddle carrier STS crane Ship-to-Shore crane TEU Twenty-foot equivalent unit xvii 1 Introduction This chapter presents the background to the issue investigated in this thesis. It also describes the purpose of the work and the research questions to be answered. 1.1 Background Container transport is the most important transportation mode for international trade, and it is considered a key for economic globalization (L. Chen, Xu, Zhang, & Zhang, 2018). The international shipping industry is currently responsible for around 90% of world trade (Grbic, 2016). The competitiveness is high and concepts like slow-steaming, when vessels reduce speed to lower fuel consumption, is com- monly used to reduce cost and increase profit (Ha & Seo, 2017; Liang, 2014). For sea transport to be operable there has to be some kind of transshipment between sea and land. These transshipments are usually managed by ports around the world where the goods are moved between trucks, railway and vessels. Ports are also im- portant for the economic activities in the hinterland since they are the connection between sea and land transport (Dwarakisha & Salima, 2015). Ports throughout the world are connected by several complex networks of long going and temporary shipping lanes (Ducruet, Rozenblat, & Zaidi, 2010; Kaluza, Kölzsch, Gastner, & Blasius, 2010; Mærsk, 2017). The main lanes of major shipping compa- nies are operated with some of the largest vessels in the world (Mærsk, 2018; MSC, 2015; OOIL Group, 2017), which are able to carry over 21 000 ’Twenty-foot Equiva- lent Units’ (TEU) (Lumsden, 2007). The size of these vessels limits which terminals are able to accommodate them due to harbor depth, size of ship-to-shore gantry cranes (STS cranes) and demand (Thoresen, 2014). This is significant to determine the delegation of vessels to different shipping lanes (Mærsk, 2018). Major ports are connected by direct lanes and feeder traffic which simply is transshipment to minor ports with smaller vessels. Feeder traffic distributes and collects goods to support the intercontinental lanes and gain economy of scale (Mærsk, 2017; Unifeeder, 2019). Most ports are the connection between sea transport and land-based transports on trucks or railway (Lumsden, 2007). The difference in capacity between vessels and land-based transportation modes are handled by either storing containers in the 1 1. Introduction terminal or large number of trucks and trains connected to the arrival of the vessel. Most common is to store containers which divide the intermodal change over an extended period of time compared to the duration when vessels are at the port. The port operator is the one who manage most of the activities in the terminal such as storing, discharging and loading. Both at quay with different vessels and land side with multiple trucks and train sets at the railway (Lumsden, 2007). Terminals are operated quite different depending on the invested infrastructure (Thoresen, 2014). Some terminals are more automated than others, using auto- mated guided vehicles (AGVs) to move containers between STS cranes and storage (Pjevcevica, Nikolica, Vidicb, & Vukadinovica, 2017). Other port operators are more manual and use straddle carriers, terminal trucks and/or reach stackers. These ma- chines are driven by operators to move containers and oversized good (Lumsden, 2007; Thoresen, 2014). AGV shuttles or straddle carriers are typical machines serving the STS cranes with containers, they are referred to as the ’fed delivery service’. STS cranes are used in most container ports to unload and load the vessels at quay. The modern STS cranes can accomplish about 30-40 containers lifts per hour if the fed delivery service is working properly. In the largest port there can be up to six STS cranes who works simultaneously with the same container vessel (Thoresen, 2014). The ones who allocate the jobs in the terminal for both straddle carriers and STS cranes are called dispatchers. When the vessel is planned a computer-software distribute the jobs but if the unexpected happens the dispatcher manually select the jobs to make sure everything will work in the end (Bish et al., 2005). Figure 1.1 depicts a side view of a container terminal with the STS cranes at the quay and container stacking straddle carriers in the yard. Figure 1.1: Side view of a container terminal adapted by Thoresen (2014). Studies show that by using higher number of AGVs when running a terminal opera- tion, the utilization of the STS cranes increases and in the same time the utilization of the AGVs decrease (Pjevcevica et al., 2017). The utilization is the extent to which space, workforce or equipment like STS cranes, AGVs and straddle carriers, are used (Jadayil, Khraisat, & Shakoor, 2017). For instance, idling, which is when machines are standing still, gives a negative impact on the utilization (Burgess, Peffers, & 2 1. Introduction Silverman, 2009). Being able to balance the utilization of the different operating machines can decrease the time it takes to turnaround a vessel in port and ensure that it departs on time (Pjevcevica et al., 2017; Zhen, Xu, Wang, & Ding, 2016). Information sharing is the transfer of data between individuals and it can poten- tially decrease the turnaround time at seaport that uses a more traditional container handling (Sonnenwald, 2006). Enhancing the workers with information through new technology could increase the ports’ performance (Romero et al., 2016). As a com- parison to an automated operator who is only able to add more vehicles to improve performance if it is truly efficiently programmed. According to Folan, Browne, and Jagdev (2007), when a firm operates and expands, the firm itself would not be able to resist, analyzing how this change are compared to present or other similar op- erations. The performance is the objective in which this comparison eventuates in (Folan et al., 2007). Mili and Gassara (2015) state in their article that even if the distance traveled for manual straddle carriers is minimized as much as possible, by for example genetic algorithms, there is still a possibility of human error. Especially if the operators find it difficult to follow the complicated itineraries assigned to them. These errors could be reduced by using a proper communication and tracking system (Mili & Gassara, 2015). Reducing the errors will give a positive operational effect. Operational effect is when a new implementation affects the operation, the implementation can have either a positive or negative effect (Duistermaat et al., 2007). 1.2 Case terminal Most part of the thesis was carried out at a case terminal were one of the authors had previously worked. The authors had access to the terminal during the whole project and this was where the observations, interviews and measurements took place. The case terminal should be thought of as a seaport and a big actor among the world’s container transports and hubs. The investigated case terminal uses the operating system Navis which is used by around 70% of the world’s terminal operators (Navis, 2018). Today’s settings might limit the machine operator which needs to be investigated further if it affects the overall performance. 1.3 Purpose The purpose of this work is to investigate how unnecessary stops of STS cranes can be reduced by improved management of straddle carriers through information sharing. The STS cranes performance are crucial for terminal operators to be able to assure turnaround times so vessels can depart on time (Pjevcevica et al., 2017; Zhen et al., 2016). Improved information sharing to the straddle carrier operators could lower the risk of unnecessary stops for STS cranes and increase opportunities for 3 1. Introduction operators to take own initiatives. The information could also allow them to make decisions that improves production rates, increases movement, reduces the stress level and intriguing the employee to actively look for things around them that can be improved or acted upon. 1.4 Research questions To fulfil the purpose of this work, three research questions need to be answered. RQ1 - How are STS cranes and straddle carriers operated in container terminals? This is done to clarify how the container and straddle carrier flows are operated around the STS cranes. What different settings and operations that are performed to discharge and load container vessels. RQ2 - What information is required to be shared to improve the performance of straddle carriers in container terminals? There might be many improvements for the straddle carriers that could emerge from shared information. It is valuable to find differentiated improvements depending on operations related to discharge, loading and preparation work etc. RQ3 - What are the operational effects of information sharing in container termi- nals? The findings will have to be evaluated. Some of them will be tested in live production to see if the outcomes support the theoretical improvements. 4 2 Review of relevant literature Within this chapter, literature related to general container terminals, the particular case terminal and framed scope of this thesis work is reviewed. 2.1 Intermodal container transports Container transport are a pristine intermodal solution, having standard units that are transferred between different modes of transportation, normally sea, rail and truck. Transfers are performed at several nodes in the transportation networks connecting different terminals and, in the end, almost any destination in the world (Lumsden, 2007; Roso, 2013). Typical intermodal transports can contain both truck and railway transports before and after a voyage on a container vessel (Roso, 2013). 2.1.1 Containers According to literature, a container is a reinforced steel box developed to allow goods to be packed and sent without any additional handling before discharge at the final destination (Babicz, 2015). Containers have corner fittings in all corners, which are the connection points for machines, locking devices and lashing equipment, in order for the container to be transported and handled (Thoresen, 2014). There are several different characteristics a container can bear. They can be ventilated, insulated, refrigerated, open top, bulk liquid, flat rack, vehicle rack or equipped with interior devices (Babicz, 2015). Common measures are (TEU) ’twenty feet equivalent unit’ and (FEU) ’forty feet equivalent unit’. TEU and FEU describes the twenty feet (20’) and forty feet (40’) containers (Babicz, 2015; Rushton, Croucher, & Baker, 2010). As an example; vessels are measured by the number of TEUs they can carry to describe a vessels maximum capacity. When containers in general differs in sizes this is used as a common equivalent measurement index (Thoresen, 2014). Containers are part of the ISO standard and sometimes described as ’ISO containers’. Within this standardization the containers are defined with measures and features to enhance the use all over the world (Rushton et al., 2010). The standard lengths are 20’ and 40’, but there are containers with lengths of 24’, 45’, 48’ and 53’. Containers 5 2. Review of relevant literature can also differ in height and width. Height is either eight feet six inches (8’6") for a normal cube (DV), also known as dry van, and 9’6" for a high cube (HC). Width is either 8’0" or 8’6" for ordinary cubes (Babicz, 2015; Rushton et al., 2010; SeaRates LTD, 2019). There are also possibilities that vehicle racks, flat racks and open top containers can be loaded with oversized goods, the first two with wider loads and all three with higher loads than the standard container heights (SeaRates LTD, 2019). How they are further handled and what machines are used to be loaded or discharged from vessels are explained in section 2.1.4 under "Oversized goods". 2.1.2 Container terminal According to Babicz (2015) a container terminal is "An area designated for the stowage of containers; usually accessible by truck, railroad and marine transporta- tion. Containers are picked up, dropped off, maintained and housed there". 2.1.3 Operations in container terminals Container terminal operations are dependent on long term investment. What is acquired today is calculated to be in production for many years, if not several decades (Bartošek & Marek, 2013). When demand changes over time, increased volume and calls of larger vessels arriving to the terminal is constrained to be serviced by these assets. Already invested infrastructure that initially were calculated might be outdated or fairly undersized. This does not mean that new investments and potential optimization would be viable before already existing infrastructure has been paid off or become obsolete (Bartošek & Marek, 2013). Terminals, old and new, are having these challenges. Not only performing on every container move, but at what level the terminal area is used, which is considered a limited resource. Even smaller terminals can change their ability to handle larger volumes through a change of infrastructure, but reasons as costs, foundation and ground support, duration of the transition and current equipment and its life span intervals can be part of that decision. 2.1.4 Equipment in container terminals In this section, basic container handling equipment is presented. This is done to later provide the reader with information on how the equipment is used in the terminal. By defining the equipment, the reader will be able to understand the work-related issues presented in ’Chapter 4’ that can affect measures and the way the authors have performed their experiment and data gathering to avoid those issues. Ship-to-shore Cranes are electric rail mounted cranes that operates along the quay, directly perpendicular to the vessels berth position (Bartošek & Marek, 2013). The cranes are able to move along the side of the vessels to operate all possible 6 2. Review of relevant literature container positions on board. Containers are then moved between the vessel and land side where it is handled by the ’fed delivery service’. According to Bartošek and Marek (2013), the first STS crane dates back to 1959, and was built in Alameda, USA. That crane was designed to handle containers weighing 23 tons, lifting 16 meters over the rail and with an outreach of 24 meters over the vessel. Today’s cranes have more than doubled in size and are capable of lifting 75 meters high and with an outreach of over 70 meters weighing up to 1800 metric tons (Bartošek & Marek, 2013). STS cranes are calculated to operate over a period of 25 years, but effective life span will not exceed 20 years. Vessels are continually getting larger and STS cranes are directly connected and needs to be able to handle these future new vessels (Bartošek & Marek, 2013). STS cranes are capable of handling 30 to 40 container lifts per hour, averaging cycles lifts between 90-120 seconds. Terminals that are equipped with straddle carriers are able to support the highest numbers with an ability to produce up to 40 lifts per hour (Thoresen, 2014). Terminals equipped with terminal trucks are only able to achieve between 28-35 lifts per hour Thoresen (2014). On the STS crane, there are two major openings that goes straight through the construction, (1) first one is in the bottom of the crane and runs parallel to the vessel, the width of this space is also referred to as the ’crane rail gauge’ (Bartošek & Marek, 2013), and (2) the opening underneath the beam that stretches over the vessel underneath and behind the STS crane, which is known as the outreach and back reach. The space below the beam runs perpendicular to the vessel and rail and allows containers to be lifted between vessel and shore. The lower opening creates a space formed as a roofless tunnel with the crane structure that limit the entrance. This space is used by the ’fed delivery service’ to operate in and through. This tunnel runs parallel to the quay and rail beneath the STS crane. The height of the entrance and width of this tunnel differs between STS cranes, much due to already invested infrastructure or machines e.g. already existing rail or chosen fed delivery service (straddle carrier or AGV shuttle carrier). Normally the rail gauge is between 16-35 meters and the minimum height is twelve meters (Thoresen, 2014). A rail gauge of 35 meters will allow six truck lanes between the STS crane legs (Thoresen, 2014). The second opening runs through the crane construction above the leg structure and perpendicular to the vessel, rail and tunnel. The opening underneath the beam is connected to the tunnel and creates an operational space, the width of this space is minimum 16 meters (Thoresen, 2014). The weight of the STS crane is resting on the rail, both at the quay side and at the land side legs. The two openings and their connection allow the crane to operate containers and hatch covers either over the vessel, under or behind the STS crane and in between. Straddle carrier is a specialized container handling vehicle that allows a terminal to stack containers three high with a minor gap between the container rows, to be able to access them. The straddle carrier is a high raised vehicle which straddles the container and carry the container in its belly (N Spasovic, Professor, Sideris, Das, & Chao, 1999). The machines are diesel electric or diesel hydraulic which powers the four center wheels and with steering on all eight (KoneCranes, 2017b). The 7 2. Review of relevant literature four front wheels turn in one direction and the four in the back turns in the other direction, which gives the vehicle extraordinary maneuverability in the terminal, both under cranes and around the yard. Depending on model, they can lift from first to a fourth layer, able to operate every individual container within a few extra moves, independent of position and length of the rows (KoneCranes, 2017b). The straddle carrier is equipped with a yoke, also known as spreader, that attaches to the containers corner fittings. The machine is able to lower twin boxes to handle twin 20’ containers. The yoke is adjustable between 20’, 30’ and 40’. All containers can be lifted with the 20’ and 40’ position and the 30’ is dedicated to move the cone handling platform, which is further explained in section 4.1.2 under platform (KoneCranes, 2017a). When the yoke is in 40’ position, twin boxes can be lowered, which enables the machine to lift and handle two 20’ containers as twins. To easier lift separated 20’ containers or just create a gap between the containers when placed in the yard e.g. when placing them on top of other already placed containers that may be separately placed. The yoke is guided by the construction and can be adjusted by hydraulic pistons to easier align and lift targeted containers. The joke is then lifted by four cables, one in each corner of the yokes fixed guiding construction. Straddle carriers are often top heavy, the construction is heavy since the engine is placed in the top of the machine together with other lifting accessories. In addition, when the spreader and container are lifted the center of gravity shifts, especially when carrying a loaded container. Operating speed is limited in transport mode and further decreased when the container is lifted higher to reduce the risk of tumbling over (KoneCranes, 2017a, 2017b). Straddle carriers are suited for smaller terminals, with minor need for reinforcement of the terminal pavement and fairly simple ways of making alterations to the terminal layout (Thoresen, 2014). Terminals with straddle carriers stack containers either two or three high. This system is normally the fastest system for a terminal that handles 100 000 to about 3 000 000 TEU per year (Kalmar, 2007; Thoresen, 2014). According to Thoresen (2014) typical generalization to a straddle carrier system are; (1) three to five straddle carriers per crane, (2) straddle carriers perform an average of 10 moves per hour, (3) the medium density in the yard is 500-750 TEU per ha, (4) the STS cranes have a high productivity with a buffer zone under the crane, (5) the system carries a high labor, capital and operation cost, (6) the system is very flexible in regard of relocation within the terminal, (7) the terminal layout is easy to alter, (8) high control, when trucks are restricted to certain receiving areas. Terminal tractor is a type of truck equipped with a fifth wheel coupling to trans- port trailers short distances mainly within industrial areas or ports (KalmarOttawa, 2018; Thoresen, 2014). This specialized vehicle can hitch trailers, flatbeds and wag- ons and raise the trailer front without raising the landing gear. Often used in differ- ent terminals and other transport businesses, moving trailers and other equipment (KalmarOttawa, 2018). Container terminals use basic wagons that are adapted to keep containers in place during the short transports within the terminal area. With small elevated lumps for the containers corner fittings that are placed along the wagon’s chassis at standard distances. 8 2. Review of relevant literature Reach stackers is another vehicle that is used to stack containers in terminal yards. It has the ability to stack containers with a high density which gives an increased yard utilization in the terminal. Reach stackers can be used to load and unload flatbeds that are moved with terminal trucks serving as the fed delivery service. A reach stacker system is suited for small terminals handling approximately 200 000-300 000 TEU per year. This system can stack 4 containers deep in stacks of six high but in many places, containers are just stacked 2 in depth and 3-4 high, to avoid large amounts of reshuffling when needed in other orders (Thoresen, 2014). According to Thoresen (2014) there are generalizations that can apply to the reach stacker system; (1) there are normally three to five terminal trucks and two reach stackers per STS crane, (2) the yard will only be utilized with about 500 TEU per ha with stacks of 4 high, (3) will contain a medium productivity for the STS crane with a non-buffer zone beneath the STS crane, (4) low capital and operations cost, but high labor costs and (5) low control, allowing trucks in the stacking area. Oversized goods are transported in vehicle racks, flat racks or open top containers (SeaRates LTD, 2019). Special over height frames are placed on the container frames working as an extension to avoid damages to the goods (Tec Container Asia Pacific, 2015). These are located in the terminal and transported to and from the vessel by the fed delivery service. These goods are often handled in a parallel flow, loaded on flatbeds pulled by terminal tractors, and loaded and unloaded with reach stackers in the terminal and by STS crane at the vessel. These goods are often loaded on board, either on top of containers, below or above deck when they constrain what is able to be loaded around or above (SeaRates LTD, 2019). Rail mounted gantry (RMG) and rubber tyre gantry (RTG) cranes are used to stack containers in the yard. From a vessel, the STS crane places the con- tainer on a shuttle carrier or a terminal tractor which moves the load to the RMG or RTG system. These two cranes drive over several rows of containers that are placed rather tight together. They can span between 20-50 meters depending on construction, but usually 5-9 containers wide, around 10x40’ containers in length and between 4-6 containers in each stack (Nidec, 2009; Thoresen, 2014). When run- ning normally the number of moves can differ between 15 to 25 containers per hour (Thoresen, 2014). According to Thoresen (2014) these systems are economically viable for terminals that handles over 200 000 TEU per year. RTG or RMG cranes are the only practical solution for terminals with either restricted or expensive land areas when handling large numbers of containers. The yard density of this system when containers are stacked four high are about 800 TEU per ha. RTG cranes in particular have been used worldwide in terminals for many years. There are several different manufacturers lowering both maintenance and capital infrastructure costs (Thoresen, 2014). According to Thoresen (2014) generalizations can be applied to RMG and RTG supported by shuttle carriers; (1) high productivity for the STS crane with buffer zone under the crane, (2) one STS crane is supported by two RT- G/RMG cranes and two to three straddle carriers, depending on distance from the yard cranes to the berth sight, (3) the system requires high capital costs, medium operating costs and low labor costs. When operated with support of terminal trac- tors. According to Thoresen (2014) generalizations apply; (1) medium productivity 9 2. Review of relevant literature for the STS cranes and no buffer zone under the cranes, (2) one STS crane is sup- ported by two RTG/RMG cranes and two to three shuttle carriers, depending on distance from the yard cranes to the berth sight, (3) the system requires medium capital and operation costs but high labor costs. RMGs are fixed installations that runs along rails and RTGs are able to move between the different yard blocks. RTGs are able to turn the wheels 90 degrees and travel perpendicular to the blocks, along the yard, to operate in different blocks depending on need and utility. These crane types can be one, two or three in combined settings in one block. If more than one, they are either two identical or one that is higher than the other, that can move over, making them switch places in the yard block. This setting is normally performed with RMGs that follows individual tracks, keeping the two cranes away from each other when passing to avoid collisions. There are also combinations with two identical cranes, with a third, larger crane, that can over pass the two others. Increasing the productivity and availability in the yard (Carlo et al., 2014). In addition to road connections, some terminals are connected with railway from the inland. These types of inter-modal connections are handled with one or more RTGs that span several rail tracks (Carlo et al., 2014). Containers are then transported to inland container terminals, often referred to as inland ports or dry ports (Carlo et al., 2014; Roso, 2013). 2.1.5 Vessel structure Over the year’s vessels have become larger and larger and miner improvements are implemented regularly in newer vessels. Many vessels are tailored to fit the specific shipping companies needs regarding ocean liners or feeders and they are often built in series of at least a few sister ships to divide and carry the development costs (Lumsden, 2007; Mærsk, 2017). Even if the vessels are individually developed, they follow some basic structures and features, which some are explained in this section (Babicz, 2015). Bays and lashing bridges, aboard a vessel, bays are a vertical division of a vessel from steam to stern, used as one part of the indication of the container stowage position together with rows and tiers. Odd bay numbers indicate a 20’ position and even bay numbers indicate 40’ positions. The numbers start from low numbers close to the steam to highest in the stern (Babicz, 2015). Some vessels are equipped with dedicated single bays for 20’ containers, mainly vessels with extended guides above deck or vessel called open hutch, that are further described in section 2.1.5. Guides above deck can hold both 20’ and 40’ containers, with the addition that 20’ twin containers are in need of cones to stay in position, avoiding sliding out of place and damage other goods (Babicz, 2015). The largest vessels today are able to stack ten containers above deck. Containers by themselves are not structurally fit to cope with the forces and motions affecting 10 2. Review of relevant literature them on the transoceanic voyages (Babicz, 2015). To be able to stack containers higher they are strapped to the vessel with lashing which is further explained in the section 2.1.5. There is a certain level that can be unleashed over the top lashed containers and to allow higher tiers, lashing bridges are built on each side of the bay hatches to simply reach higher lashing points and increase the tiers in each stack. Lashing bridges are strong steal structures built on deck raising from a few meters to several stories. The distance to the lashing bridge from the locked position of the containers are restricted by a minimum distance to be able to perform the lashing (Babicz, 2015). There is certain level that needs to be exceeded to stack longer containers e.g. 45’ containers, creating a distance from the lashing bridges and enable lashing in bottom of the container or in layers below. Hatch covers and holds, containers on board vessels are stacked both above deck and below in holds. To maintain a weather proof volume and to ensure a safe journey, the vessel is sealed with hatch covers dividing the load above and below deck, which is illustrated in figure 2.1. Hatch covers distribute the weight of the load to spread the forces throughout the vessels construction and prevent water to enter the holds below deck. On the larger vessels hatch covers can weigh more than 20 metric tons and be structurally strong enough to carry several hundred metric tons of cargo (Babicz, 2015). Figure 2.1: Blocked container vessel abstraction (Wilson & Roach, 2000). There are a few types of hatch covers that are used on different vessels. Some are mounted on hinges and lifted by hydraulic cylinders and maneuvered by the deck men on board. Others are lifted off the vessel and are placed on the docks concrete paving, either between the legs of the STS crane or behind the STS crane (Babicz, 2015). In rare cases hatch covers are placed above another hatch cover on board, mainly on larger vessels where there are multiple hatch covers in the same bay. Hydraulic systems are generally used on smaller vessels with one hatch cover in each bay (Babicz, 2015). A specific type of container vessel goes under the name of ’open-top’ or ’open-hatch’. This type of vessel, in all or some of the bays, have robust guides built above the normal deck line and lack hatch covers over the holds. In general, this design is common on feeder vessels and are used on vessels designed to carry up to approx- imately 4000 TEUs (Babicz, 2015). These vessels are designed with large pumps 11 2. Review of relevant literature that are dimensioned to handle excessive amount of water during the worst severe storms and weather conditions at sea, that risk to flood the open holds (Babicz, 2015). This type of vessel is designed to minimize the use of lashing equipment in an attempt to reduce the turnaround time at the terminal (Babicz, 2015). Figure 2.2: Stowage arrangement for a container vessel, adapted, by Wilson and Roach (2000) The holds are dimensioned to carry 40’ containers that are fixed within guides along the fore and aft of the bay. By fixating the containers, the need for lashing or cones are eliminated (Babicz, 2015). Each row or cell have their own guides. The heaviest goods are often loaded in 20’ containers. Two 20’ containers placed in twin are among the heaviest units that covers a 40’ footprint lifted aboard the vessel. Combining them in to a twin and placing them in the bottom layers in the holds aboard is standard procedure to guaranty the vessels stability. This leaves the center of the twin to be able to slide out of position if not fixated by cones, these cones are further explained in section 2.1.5. The heaviest containers are loaded in the bottom and units with lower weight are placed higher up in the stacks. Placing 40’ containers above twin 20’ containers is a common procedure, but the other way around is prohibited when the 40’ containers are not structurally designed to carry that load in the center of the container. The stowage plan is showed in figure 2.2, which is an illustration of a vessel’s available stowage positions, with standardized numbering and logic for all container positions (Babicz, 2015). Vessel equipment, to perform the discharge and loading sequence at port the vessel brings their own equipment, mainly ’cones’ and ’cone bins’ to be used by terminal workers, but also ’lashing rods’ and associated ’turnbuckles’. Every vessel got a limit to what number of containers they can carry, and the equipment is by that reason limited by the possible combinations of stowage (Babicz, 2015). Cones are locking mechanisms that are placed in the bottom corner fittings of con- tainers (Babicz, 2015; Thoresen, 2014). This is done either at the cone fitting plat- form or in special sequences or conditions on board the vessel. There are a few different types of cones, some prevent the containers from sliding sideways and some securely lock them to the container below. In cases where cones with locking mech- 12 2. Review of relevant literature anism are used, either deck men or terminal workers need to unlock them before the STS crane can discharge the containers (Babicz, 2015). If the containers are stacked high, they are lashed to the deck or lashing bridges with ’lashing rods’ and turnbuckles. The ’lashing rod’ is placed in the ’corner fittings’ of the containers at certain levels to secure the stacks on deck. This can differ between different vessels and how the containers are stacked. The turnbuckles are either secured to the lashing bridge, hatch covers or the deck. ’Lashing rods’ are lose metal rods with a few lumps on the end and a hinged hook in the other end. The rods are positioned in the corner fittings of containers and placed in an angle that locks the container. These lashing rods are aligned with the turnbuckle and the turnbuckle is attached to a suitable lump and screwed down to a tight fit. Unused lashing rods are stored either upon the lashing bridge or on deck (Babicz, 2015). 2.2 Yard management Terminals today are operated with different equipment, there are three widely spread types of systems in use to stack and store containers; (1) a forklift and reach stacker system, (2) a straddle carrier system, (3) a ’rubber tyre gantry’ (RTG) and/or ’rail mounted gantry’ (RMG) system, (4) combinations or mixtures of these three (Thoresen, 2014). Depending on what equipment is used there are unique options to optimize and run the terminal efficient. There are many ways to measure the efficiency of storage systems, one way is to compare the number of TEUs stored per hectare which can be seen in figure 2.3. Figure 2.3: Note: actual storage capacity depends on operational aspects such as required selectivity etc. (Kalmar, 2007). To generalize yard layouts, they consist of multiple rectangular blocks which are served by either one or several material handling machines (Carlo et al., 2014). The material handling machines are the decision basis for what type of strategy the indi- vidual terminals yard management team are using. In a study where 114 terminals 13 2. Review of relevant literature where represented, 63.2 % of the terminals used RTGs as their main material han- dling equipment. Making RTGs the number one main handling system, followed by straddle carriers that in the same study where the main material handling equipment in 20.2 % of the terminals according to Carlo et al. (2014). Figure 2.4: Asian (a) and European (b) storage yard layout (Carlo et al., 2014). The utilization of RTGs and RMGs can differ between terminals. Partially depend- ing on the different types of fed delivery service that is used in the terminal and the level of control e.g. if external truck drivers are allowed in to the yard area or not (Carlo et al., 2014; Thoresen, 2014). As seen in figure 2.4 the layout and stor- age under the cranes and the input/output (I/O) points, where the containers are loaded and unloaded differ from long-side in example (a) and short-side in example (b). This doesn’t only affect the number of possible stored containers under the crane but also how the terminal wants to store and stack containers in the yard. Utilizing storage type (a) showed in figure 2.4, the fed delivery service parks close to the container or its planned position beneath the gantry crane to be lifted. This saves time when the lifting distance to and from the vehicle is reduced, but the trav- eling distance of the vehicle is longer. Utilizing storage type (b) showed in figure 2.4, makes the distance from and to the berth place shorter for each vehicle and an increased storage beneath each gantry crane. This alternative provides longer traveling distance for the crane and preferred option of having the container position close to the I/O point in close connection to the berth if it is an export container and closer to the land side I/O point if it is an import container. Containers are then moved within the block to prepare imports and exports for later transition. As mentioned in section 2.1.4 under "Rail mounted gantry (RMG) and rubber tyre gantry cranes (RTG)", the blocks can be served with more than one RTG or RMG crane at once. This increases the ability to re-handle and prepare the blocks to reduce time when containers are ready for a move to be loaded on vessels or trucks (Carlo et al., 2014). 14 2. Review of relevant literature Figure 2.5: Storage yard layout for straddle carrier (Carlo et al., 2014). Terminals that uses straddle carriers as their main handling system are also using several rectangular blocks as storage areas, with the miner difference of having larger gaps between the container rows. Normally a distance of half a container width is used, making the blocks generally 50 % wider compared to when the containers are stacked close together. A straddle carrier yard layout is roughly illustrated in figure 2.5 with machines and gaps. These gaps are required for the machine to be able to straddle and move within the row to stack containers. The rows are referred to as linear stacking. Figure 2.5 shows the rows parallel to the quay but there are blocks in straddle carrier terminals that are perpendicular to the quays (Carlo et al., 2014). All different systems used in ports requires terminal operators to stack containers on top of each other and in specific places and orders, where in the end, there is a need of a re-positioning. Either to make place for other containers or to move the containers closer to the I/O points to reduce the time it takes to perform the planned moves during the call of the vessel. There are also types of re-handling where containers are gathered in the same area, where they are either going to the same port or they are specifically placed in sequences to be loaded in a certain order on the vessel. These types of moves are referred to as re-marshalling or housekeeping (Carlo et al., 2014). Terminals can be divided into a primary and a secondary yard area. The primary area considers the storage area for loaded containers in direct contact with the berth area, where the STS cranes are active (Thoresen, 2014). The secondary yard includes the empty flow with storage area, container maintenance, repair areas and equipment storage e.g. extension frames, lashing cages for manual work on unreachable heights and chains to lift damaged containers (Thoresen, 2014). 15 2. Review of relevant literature 2.3 Information sharing Information is data that can lead to less uncertainty and increased understanding. It is timely, accurate and presented in a relevant and meaningful way. Information is important since it can influence actions, decisions or results taken by different actors (Losee, 1998). Information sharing is the transfer of data between people, organizations and tech- nologies. Sharing information, either upon request or proactively, changes a person’s view of the world and creates a mutually compatible understanding of the environ- ment (Sonnenwald, 2006). It includes supplying information, confirming that the information is understood and that it has been received. Information sharing is an important activity when people collaborate with each other. When a group works together, there must be a continual information flow between the members and an ability to understand the information that others provide. If information is not shared in a good and effective way, the collaborative work fails (Sonnenwald, 2006). According to Mohr and Nevin (1990), there are four fundamental parts when com- municating information: frequency, direction, modality and content. Frequency refers to the amount of shared information between actors in the organization. The frequency can be both big and small. A minimum amount of shared information is crucial but sharing too much information can give negative effects. The direction describes if the information is shared vertically or horizontally among the actors in the organization. Generally, information is shared downwards from the more powerful actor to the weaker actor. The direction can also be either unidirectional or bi-directional, e.g. information is shared in one direction or two directions. The modality specifies the method which is used to share the information. It can be done by writing, telephoning, speaking face-to-face or in other ways. Modality can also be defined as if the information sharing is structured and adjusted or spontaneous and non-adjusted. The final part is content which refers to the kind of information that is shared. Two frequent ways of categorizing content is the type of shared information and the type of influence strategy in the shared information (Mohr & Nevin, 1990). Heilig and Voß (2017) state in their article that sharing information through dif- ferent systems have become essential for the competitiveness of ports. It enhances the decision making which gives increased reliability, productivity and security in ports. Real-time information and data have increased in importance to improve the planning and coordination of activities among actors in port operations. The positioning data of objects, such as containers, are necessary for forecasting which is a foundation for long- and short-term decision making. The importance of shared information will increase in the future due to challenges and competitiveness faced by ports around the world (Heilig & Voß, 2017). 16 2. Review of relevant literature 2.4 ARA-model The ARA-model contribute with a theoretical framework of the operation and results of interaction. It is mainly used to describe the interaction between companies in business relationships. A relationship is a jointly cooperation between mutually dedicated groups. The model contains three different layers: Actor Bonds, Activity Links and Resource Ties (Håkansson & Snehota, 1995). The model suggests that each of the three layers are closely related and affects each other (Ford, Gadde, Håkansson, Snehota, & Waluszewski, 2010). Figure 2.6: The ARA-model with its three layers (Håkansson & Snehota, 1995) 2.4.1 Actors Individuals are the ones that affect what happens in networks. They do not act by themselves, instead they interact with each other and their actions becomes system- atized. The actor dimension is restricted by the activity and resource dimension. In some industries changes in connection and actor design might take several years due to inactivity and resource status. The actor layer goes beyond the other two and it is more complex than the other with more soft and abstract values (Håkansson & Snehota, 1995). The layer is constructed on how much the actors know see and feel close to one another. How they like, impact and trust each other and evolve a dedication together. Connections between actors differs in strength which influ- ences how the involved individuals see possibilities and achievable directions for the interaction (Ford et al., 2010). 2.4.2 Resources Actors use resources which can be in the form of financial, personal, technical or other shape. Resources can be tangible such as physical items like equipment or a factory and intangible like knowledge. There is no actor that has all the required resources, so resources are shared between actors. The resource layer describe how resources are used between actors when their interaction develops. Sharing resource between each other can increase the degree of utilization of each resource (Ford et al., 2010). Studies show that actors develop resources, new products and use 17 2. Review of relevant literature products in new ways. These improvements of resources often advance from rela- tionships between actors since when sharing something together increases the chance of confronting how a resource is produced (Håkansson & Snehota, 1995). 2.4.3 Activities Activities are carried out by actors, there are different activities like buy and sell, develop, logistics, information handling and production. A large number of activities are performed by actors. The way that the activities are accomplished regulates the revenues and costs for actors. It is not only important to do things right, but actors should also do right things. Several activities are not carried out in isolation instead the activities performed by different actors are connected and dependent on each other (Håkansson & Snehota, 1995). The actor’s activity structures can become more or less linked and integrated. It has been shown that the stability and strength of the activity links have a great impact on the financial effects for the concerned actors (Ford et al., 2010). 2.5 Lean Lean originates from the Japanese company; Toyota and it is a method used in manufacturing processes to minimize and eliminate waste. The lean technique con- tains several different principles which are all important to know before adopting the method (SixSigma, 2017). According to SixSigma (2017), the first principle is, as mentioned before, to eliminate waste. When eliminating waste, companies should examine all different areas in the system and detect the work that does not contribute with any value. Everything in the manufacturing operation that does not add value to the products is considered a waste. In Toyota, seven different wastes were found which can be seen below (Berlec & Starbek, 2019). 1. Overproduction 2. Waiting 3. Transport 4. Inappropriate Processing 5. Unnecessary Inventory 6. Unnecessary Motion 7. Manufacturing Defects 18 2. Review of relevant literature Another principle is that companies should aim for having a leveled production, which means that the workload should be evenly distributed over all working days. The risk of overproducing is significantly increased if a company increase the pro- duction effort when they get a large volume order. The most important principle in lean methodology is to have continuous improvements. A company will not have any progress at all if they do not improve. But it does not matter if the organization have small or big improvements, the important thing is that the company is open for new enhancements, which in the end will lead to progress (SixSigma, 2017). Liker (2004) summarizes Toyotas lean thinking with the 4P model (see figure 2.7). The first part of the 4P model is Philosophy, a company should base their de- cisions on long-term thinking even though it might lead to costs in the short-term. The second part is Process, if the leaders adopt the right process, they get the right results. They should create process flows that unveil problems and implement standardized work as a foundation for continuous improvements and the workers par- ticipation. Another important part in lean is the People and partners, companies should evolve leaders that understand the work, practice the company’s philosophy and learn it to others. If a person wants to grow personally and has the capacity to do it then they are to invest in. The last and final part is Problem solving, lean encourage people to go and see the problem themselves to really understand the situation and by that take the right decision. Companies should become a learning organization by continuously reflecting and constantly improving (Liker, 2004). Figure 2.7: 4P model (Liker, 2004). 2.5.1 Lean in container terminals According to Christopher (2000), lean works best when the demand is predictable, relatively stable, and the volumes are high. A container terminal differs from man- ufacturing since it does not produce anything and terminals handles containers in two flows, both export and import. Vessels could also be loaded and discharged at 19 2. Review of relevant literature the same time which requires high coordination and has led to a more complex pro- cess (Casaca & Bernard Marlow, 2003). Therefore, lean needs some modifications before implementing it at a container terminal since it is primary constructed for the manufacturing industry (Olesen et al., 2015). Olesen et al. (2015) recommends a framework which can be used when implementing lean at container terminals, the framework consists of four fundamental principles which can be used to enhance the physical flow in terminals. The framework can be seen in figure 2.8. Figure 2.8: Framework which can be used when implementing lean at container terminals (Olesen et al., 2015). The first principle is waste elimination. Olesen et al. (2015) declare that the concept of waste elimination has to be changed since some of the wastes in lean are hard to translate when considering terminals. For instance, transportation is considered a waste in lean, but this is the main reason for why terminals exist. Another waste is inventory, but since terminals are paid for keeping inventory, it should be seen as a value adding service instead of waste (Olesen et al., 2015). The second principle is standardization. According to Olesen et al. (2015), one of the main reasons for using standardization is to minimize the variation in processes and to encourage continuous improvement. It is suggested to use standard work when trying to eliminate variation in operations at container terminals (Shingo, 1989). One important part in standardized work is visualization. Normal work procedures should always be at the workplace and up to date. These work procedures are called standard operations and procedures (SOPs). By instantly visualize crucial information the risk for confusion and time for trying to find the information is reduced (Bicheno & Holweg, 2009). 20 2. Review of relevant literature The third principle is leveling. Jones (2006) states in his article that leveling is used for smoothing the variation between internal operations. It is used in lean with the purpose to establish a more coordinated flow and to deal with the variability of vessel arrivals. Variability between internal processes leads to irregularities in activities and an overburdening of workers. By using the depot function more in a terminal the loading and unloading could be leveled through preparations before peak demand (Olesen et al., 2015). The fourth and last principle is continuous improvement. The concept has been used successfully at plenty of organization to encourage workers to continuously look for improvements in their daily operations (Imai, 1986). By using basic activity mapping tools, a terminal could reach proper countermeasures which can improve operations and increase value-adding processes (Olesen et al., 2015). 2.5.2 Measure activity time According to Rubin and Babbie (2009), it is not simple to find and eliminate waste and measuring the difference between various solutions could also be problematic. One way to measure the time of activities is to let the machine operators do self- observations. However, measuring activities with self-observation could lead to bi- ased data both by changing the data and also by working in a different way than usually just to improve the data (Rubin & Babbie, 2009). The machine operators’ tasks can also require a lot of focus, which can imply that they do not have time to register everything which results in lack of data, especially when data is gathered by hand. To contribute to increased motivation and thereby better data, one can use apps for phones to measure times by pressing button. StarDriver is a smartphone- app with the task to measure activities of truck drivers. It is built on the lean framework to find and eliminate time waste (Prockl & Sternberg, 2015). Machines operators might also do multiple activities at the same time which lead to a risk of increased confusion when measuring different activities. An option to increase the details in the data is to measure with participant observations. The quality of data improves but it is costly and if much data is needed it is not feasible. Also, if the sample is relatively small the participants who gathers the data are reserved with reporting special situations since it could threaten the anonymity of the operators. It is, however, apparent that the data quality is better when measured by participant observations compared to self-observations (Prockl & Sternberg, 2015). 2.6 Statistical testing Statistical testing is used to give meaning of data series, it involves planning, data collection, analyzing and concluding (Ali & Bhaskar, 2016). Statistical hypothesis testing is popular when an analyst wants to define a substantive claim (Veazie, 2015). A hypothesis is a theory or a suggestion which is used for explaining why 21 2. Review of relevant literature an observed phenomenon happened (S. Dubois, 2017). If a scientific hypothesis is proven correct in repeatable experiments, it could become a theory or even a law of nature. Hypothesis testing is commonly used to make decisions from gathered data. Simply put, a hypothesis test concludes if an observed phenomenon is likely to really reflect the reality based on statistics. According to S. Dubois (2017), hypothesis testing is one of the most essential parts in statistics since it determines if something really occurred. S. Dubois (2017) states in her article that before testing the observed phenomena, a hypothesis or guess of what could happen has to be made. The hypothesis could be that certain groups differ from each other or that an improvement in a process has a measurable outcome. By establishing your hypothesis there are two possible outcomes, the first one is the null hypothesis that there was no difference or that the improvement did not have any effect and the second is the alternative hypothesis which means that the guess is correct. In other words, when you are testing a hy- pothesis, you attempt to identify if something occurred and are comparing against the likelihood that nothing occurred. This is pretty confusing since you are at- tempting to disprove that nothing occurred. By disproving that nothing occurred, the conclusion that something occurred can be made. You draw your conclusions when all the data is collected, and you tested the hypothesis against the risk for chance. If the null hypothesis is rejected, you are claiming that the outcome did not occur by luck and that it is statistically significant. By rejecting the null hypothesis, the alternative hypothesis is accepted. If the null hypothesis cannot be rejected the conclusion is that there is no difference in the study (S. Dubois, 2017). Significance levels are used to show how likely the result from your data is due to chance (du Prel, Hommel, Röhrig, & Blettner, 2009). Significance limits are generally stated in advance to allow a determination between the null hypothesis and the alternative hypothesis. The most common used significance level is 0.05 (or 5%), at this point the result is considered to be good enough to be believed. This significance level also minimizes the risk that the null hypothesis is wrongly rejected. When the significance level is 0.05 the risk that the null hypothesis is wrongly rejected is 5% (du Prel et al., 2009). The result of a statistical test is the p-value. The p-value is the probability that the null hypothesis is false. If the p-value is less than the significance level the result is considered significant, the null hypothesis is rejected and the alternative hypothesis is accepted which means that there is a difference (du Prel et al., 2009). 2.6.1 Boxplot According to Portela, Ribeiro, and Gama (2019), an outlier is an extreme value in data series which lies far from other data points. Outliers are normally regarded as unusual values affecting the overall test due to their extreme values and they should therefore be removed. Outliers could originate from errors in measurements which might lead to over- or underestimates of studies (Portela et al., 2019). 22 2. Review of relevant literature One of the most common methods for detecting outliers is a boxplot analysis (Portela et al., 2019). It is a simple graphing tool that provides a lot of details about the distribution in data series (Krzywinski & Altman, 2014). An example of a boxplot can be seen in figure 2.9. Figure 2.9: Example over a boxplot (Galarnyk, 2018). The first thing to identify in a boxplot is the median (Q2) which is the number in the middle of the data sheet. After the median is found two interquartiles are identified. These interquartiles are in the middle of each half so the first quartile (Q1) is in 25% of the data sheet and the third quartile (Q3) is in 75% of the data sheet (Galarnyk, 2018). The interquartile range (IQR) is the range between these two interquartiles. The observations that fall outside of 1,5*IQR below the first quartile or above the third quartile are considered outliers and the once that fall farther than 3*IQR are considered as extreme outliers (Portela et al., 2019). The formula for extreme outliers can be seen in equation 2.1. Extreme outliers = Q1− 3 ∗ IQR or Q3 + 3 ∗ IQR (2.1) 2.6.2 Paired z-test One popular test used by researchers is the z-test and it is best used when the sample size is larger than 30 (J. Chen, 2019). The paired z-test is used to analyze if the difference between two population means is lower, greater or not equal to zero (NCSS, 2019). This gives three possible hypotheses. 23 2. Review of relevant literature (1) - The two tailed hypothesis test is defined as: Null hypothesis Alternative hypothesis H0 = 0 HA 6= 0 Rejecting the null hypothesis indicates that the mean paired difference is not equal to zero (NCSS, 2019). (2) - The other two are one-sided tests. The hypothesis for the lower-tailed test are: Null hypothesis Alternative hypothesis H0 ≥ 0 HA < 0 By rejecting the null hypothesis, the mean paired difference is less than zero, hence the difference between the sample means is less than zero (NCSS, 2019). (3) - Vice versa applies for the upper-tailed test: Null hypothesis Alternative hypothesis H0 ≤ 0 HA > 0 Rejecting the null hypothesis implies that the mean paired difference is greater than zero (NCSS, 2019). The random sample points are defined as xi and a paired z-test assumes that these random sample points are from a normally-distributed population and all of them has the same mean and variance. The equation used when calculating the paired z-test is given in equation 2.2 (NCSS, 2019). Z = √ n ∗ (x) σ (2.2) n - is the sample size. x - is the average value of the paired differences. σ - is the standard deviation of the paired differences (NCSS, 2019). According to Hargrave (2019), standard deviation measures the dispersion of a set of data relative to the its average value. If the sample points are closer to the mean the deviation is lower, hence the denser data the lower is the standard deviation (Hargrave, 2019). To be able to reject the null hypothesis at a significance level of 5%, the z-value has to be higher than 1.96 or lower than -1.96 for a two tailed hypothesis test. For a lower-tailed test the z-value has to be less than -1.645 and for an upper-tailed test the z-value has to be greater than 1.645 for the null hypothesis to be rejected at a significance level of 5% (Sullivan, 2017). 24 3 Methodology In this section the method that is used to fulfill purpose and research questions of this thesis will be described. The methodology includes data collection methods, workflows and design of experiments. 3.1 Research approach According to Spens and Kovács (2006), there are three different research approaches: inductive, deductive and abductive which can be seen in figure 3.1. Inductive and deductive are two general approaches, the inductive approach starts with observa- tions and then develops theory about the investigated phenomena. The deductive approach test theory by first establishing a theoretical framework and then see if it works in reality. The third approach is the abductive research approach, it starts with an observation of a real phenomenon but at the same time the theoretical frame- work plays an important role. The researcher uses a creative repetitive method of matching theory to the observation (Spens & Kovács, 2006). Abductive approach is when researchers goes back and forth between theory and observations and conse- quently increase their understanding of the phenomena (A. Dubois & Gadde, 2002). Figure 3.1: Deductive, inductive and abductive approach (Spens & Kovács, 2006) 25 3. Methodology In this thesis an abductive approach was used, it was best suited for this research since both theoretical framework and observations contributed to the analysis and conclusions of this project. First an observation that the straddle carrier opera- tors had insufficient information was made by the authors. The observation was then matched with theory about the problem. During the thesis, observations and interviews were made simultaneously with and confirmed by the theoretical frame- work. After the observations and theoretical framework was made, a hypothesis was proposed and later evaluated. 3.2 Literature review Literature reviews are performed in academic studies to get a deeper understanding of certain topics (Maxwell, 2012). In this thesis a literature review of studies and theories that were particularly relevant to the research has been investigated. Several articles, books and scientific research were studied during the process of this thesis. The literature that was examined in the review was mainly gathered from Chalmers library database, ScienceDirect and Google Scholar. Some articles which were used in the review were also recommended by the supervisor for this thesis. Important literature areas were yard management, transportation management, information sharing, lean, physical distribution management and statistical tests. The aim of the literature review was to increase the knowledge and to find a base for both author and reader. Most of the searches for important literature were made in English but some Swedish articles and books were also used in the research. Bell (2011) state in her book that it is important to eliminate irrelevant material in the literature study. Therefore, the authors were restrictive with using literature that was older than 20 years since it seldom applied to the current terminal and were outdated when it came to potential information improvements. However, one exception was the lean concept since most of the fundamental articles and books about lean were older than 20 years. 3.3 Research process This thesis was divided into three phases, one phase for each research question. In every phase a literature review was performed to increase the authors knowledge about the topic around the issues. Whenever an observation was made during a phase it was checked with theory to increase the understanding of the phenomena. In the initial phase, the first research question was evaluated. Here the goal was to clarify the flows in a container terminal to support the other research questions. To answer this question, observations, literature studies and interviews were used. The ARA model, as mentioned earlier in section 2.4, was used as a tool to facilitate the mapping of the processes. The mapping was assisted by the fact that the authors 26 3. Methodology already had basic knowledge of terminal equipment through their studies within supply chain management and previous work in the case terminal. During the second phase, improvements for straddle carriers through increased infor- mation sharing was evaluated. In order to find possible improvements, the authors’ previous knowledge was used together with observations, literature studies and in- terviews in the case terminal. The interviewees’ long work experience as straddle carrier operators was the main key when solving this research question. Several ideas were discussed with the operators who also came with suggestions on how their work situation could be improved by increased information. In the third phase, measurements were made in the case terminal. The measure- ments evaluated the information sharing’s impact on the operations in the terminal. To analyze the effect of the shared information, a hypothesis test was. Discussions were also held with the straddle carrier operators who participated during the tests to find out what they thought about the shared information. 3.4 Data Collection Several data collection methods were used to get an understanding of the case ter- minal, knowledge about terminal systems and to evaluate potential improvements etc. The primary data that were used to support the thesis were measurements, observations and interviews. 3.4.1 Observations and interviews Observations and interviews were conducted to analyze how straddle carriers and STS cranes operates and how straddle carriers assist the STS cranes in terminals. Before any observations were made at the case terminal, the case terminals website and other articles about the terminal was studied to get an understanding of the terminals function and what it handles. Since one of the authors worked at the case terminal before the thesis, basic knowledge about the terminal already existed. During the thesis, several observations were made by accompanying STS crane and straddle carrier operators. Dispatchers at the case terminal were also observed and discussions were held with them to get a deeper understanding of the movement of straddle carriers and the structure at the terminal. A table of all the performed observations can be seen in table 3.1. 27 3. Methodology Table 3.1: All observations during the project. Organization Approach Aim Case Terminal Worked as a straddle carrier operator. Before and during the time of the thesis one of the authors worked as a straddle carrier operator. Ride with a straddle carrier operator. Get a view of the activities for straddle carrier who serves the STS crane. Ride with an STS crane operator. Get an understanding of how vessels are loaded and how straddle carriers are affecting STS cranes. Study the dispatcher. Get an understanding of dispatcher’s role in the terminal. Ride with an STS crane operator. Get an understanding of how vessels are unloaded and how straddle carriers affect STS cranes. Interviews were conducted with straddle carrier operators, STS crane operators and managers at the case terminal. The interviewed managers had experience of the current software system and earlier software system used in straddle carriers. The interviews were done face to face and provided an insight in and understanding of the current situation for the operators. All the conducted interviews can be seen in table 3.2. The length of the interviews were around 30 minutes to one hour, notes were taken during the interviews and all of them were recorded so that they could be examined again if something was forgotten or if anything else was needed. Bell (2011) state in her book that when interviewing respondents one at a time it is beneficial to record the interview. This allows the interviewer to pay full attention at what the respondent is saying and afterwards ensure that the notes are correct. Table 3.2: All interviews during the project. Organization Role Aim Case Terminal SC operator 1 Understand the terminal and see what information the operators need.SC operator 2 SC operator 3 STS crane operator 1 Understand the terminal and find out how SCs affect the STS crane.STS crane operator 2 STS crane operator 3 Management 1 Find out reasons for current software and what the plans are for the future. Management 2 See which kind of data that can be gathered and the plans for the future. Saunders, Lewis, and Thornhill (2016) describe three ways in which an interview can be constructed and those are: structured, semi-structured and unstructured. 28 3. Methodology All of the interviews in this thesis were conducted in a semi-structured manner. A general guideline was used for the interviews which can be seen in appendix A, this allowed that focus could be kept on the topic of the research. By conducting semi- structured interviewees, the possibilities increased to learn more about opinions and experiences from the interviewee through asking questions that depended on the answers from the participants. The participants for the interviews were chosen due to their working experience and identified by both the authors and other involved actors at the case terminal. The interviewee agreed on that their names were not be mentioned in the thesis, this minimized the risk that any personal information would become public which could have hurt the participants. This also decreased the risk that the interviewee would restrain themselves during the interviews. Before the interview took place, the participants were informed about the reason for performing the interviews. This was done to motivate the interviewee since they sometimes do not see the benefit of answering the questions (Patel & Davidson, 2011). The result from the interviews were used in the findings and the analysis in this thesis. The interviews supported the work with the knowledge and expertise that the operators and managers possess. 3.4.2 Measurements Two different tests were performed in this project. The first test analyzed whether extra shared information could improve the container handling in the case terminal. The second test was established to see if anything else impacted the test results in the first test. The two different tests can be seen in table 3.3. Table 3.3: Table of the two different tests which were performed during the project. First half hour Second half hour First test Without information With information Second test Without information Without information To measure the difference between additional shared information and normal oper- ations, a data collection software for smartphones called StarDriver was used. The software was possible to program depending on which activities that were being measured. The software is developed by Prockl and Sternberg and originally used to measure activities of truck drivers. With few alterations it was adjusted to mea- sure activities of STS cranes and provide data of the crane operations. An image of how the app looked when used during the measurements is presented in figure 3.2. 29 3. Methodology Figure 3.2: A view of the StarDrvier interface used during the measurements. When pushing a button, StarDriver started to record the time for that activity. An example of data which was gathered from StarDriver by pushing these buttons can be seen in appendix B. N.B. the duration for each motion in the example is randomized between 40 to 60 seconds. The key performance indicators (KPIs) that were measured during the tests can be seen below: • The time for each loading cycle performed by the STS crane. • The time which STS cranes are disturbed by straddle carriers. • The distance traveled for the straddle carriers serving the crane. The KPIs were used when measuring the difference between the potential improve- ments and the regular system during the two tests. The tests were performed at four big port calls during the spring of 2019. The capacity of the vessels was around 21 000 TEUs and each port call lasted for about 72 hours. Both tests in table 3.3 consisted of smaller samples, each sample took 50 minutes and to have as similar circumstances as possible the measurements were split into parts of 25 minutes each. In the first test the straddle carrier operators worked as usual during the first 25 minutes and during the second 25 minutes they had access to additional information. In the second test the straddle carriers drove as usual during the full 50 minutes. The first 25 minutes are referred to as the first half hour and the second 25 minutes are referred to as the second half hour in this thesis. The reason for only measuring for 50 minutes was because the workers at the case terminal were following two-hour rotations and every hour operators were relieved to go on brake. The change took around five minutes and was performed each 30 3. Methodology hour. To minimize the risk that the machine change impacted the tests the time for measuring was reduced to 50 minutes instead of one hour. This ensured that the same STS crane and straddle carrier operators were operative during the full 50 minutes. Another requirement for the tests was that it had to be enough volume of containers so that the STS crane could operate continuously for 50 minutes without having to switch to another bay or be disturbed in any other way. It also had to be three straddle carriers that served the crane for the entire 50 minutes. The dispatches could regulate this depending on demand and if there were not three straddle carriers at the crane throughout the test period, that period was excluded from the data. Before the tests began, the information sharing system was tested. One of the authors drove around in a straddle carrier and the other author sent out messages to this straddle carrier. By doing this, the authors ensured that the information sharing system would work and that it could be applied in the case terminal. In addition, the minor test included sending messages both during discharge and loading. The container’s location in the terminal could only be foreseen during loading and not during discharge. During discharge the container location was calculated first when the container was picked up by the straddle carrier. Therefore, the tests were made when the STS crane was loading the vessels. During the first test one of the authors accompanied an STS crane operator and measured the crane operation times with StarDriver. At the same time, the other author sat in the dispatcher’s office and sent out messages through a computer to the straddle carrier operators who were serving the STS crane. The straddle carrier had a screen where they received their job orders. This screen had the feature to receive messages from the dispatcher with maximum 38 characters. This message function was used when sending out additional information during the first test. The information presented to the operators was carefully selected according to what had been said during the interviews. Discussions were held afterwards with operators who participated during to evaluate the additional information. The second test was made to clarify that nothing else could have affected the first test. During this test, the authors accompanied one crane operator each and mea- sured every operation for 50 minutes to see if there was any difference between the first 25 minutes and the second 25 minutes. During these samples, the crane also loaded under deck to obtain similar conditions as during the first test. The time for each loading cycle was used to see if there was a difference between the first and second half hour. All the data was adjusted to only show percentage since the case terminal did not want to share real data. For the loading cycle data, the median of the measurements during the first half hour was set to one in both tests and then all the other data points in that test was adjusted to that number. Another part that was consider during the tests was the last KPI which was the distance that the straddle carriers traveled during the samples. This data was also modified so it did not mirror the reality. The average distance traveled for each straddle carrier during the whole test was reflected upon. This data was gathered 31 3. Methodology directly from the case terminal’s operating system (Navis). Origin of the data mea- sured in the thesis can be seen in table 3.4. Table 3.4: Data origin of the measured KPI’s in the thesis KPI Data origin Time for each STS crane cycle StarDriver Disturbance caused by straddle carrier StarDriver Distance for straddle carriers Navis The data which was collected with StarDriver was stored in Excel-sheets. Excel is a spreadsheet program which can organize and manipulate data, it is created by Microsoft Corporation (Rouse, 2007). An example of data that was gathered from StarDriver can be seen in appendix B. 3.5 Statistical test Statistical tests were made to investigate whether the result obtained was due to chance or not. A programming software called Matlab was used when performing the statistical tests. Matlab is a programming language and numerical analysis environment Rouse (2015). The code which was written to analyze the data from the measurements can be seen in appendix C. The data for each crane cycle in the Excel-sheets were loaded into Matlab. After that, boxplots were made by performing the boxplot-command in Matlab on the loaded data. The boxplots were used to visualize the spread of the data. The quartiles and IQR which are used when defining the extreme outliers were calculated in the Matlab-code. The max values for each half hour was calculated in Matlab and if these max values were greater than the value for the extreme outliers that data point was manually removed in the Excel-sheet from StarDriver. This process continued till the max value was lower than the value for extreme outliers. By excluding the outliers, the risk that rare occasions would impact the result was decreased. Only extreme outliers above the upper limit was considered since there was no possibility that there would have been any extreme outliers in the lower end, assuming that no mistakes were made during the measurements. One important thing to notice is that when one extreme outlier was removed the corresponding value during the other half hour was also removed since there had to be the same number of measurement points for both half hours during every full hour. Otherwise, some of the measurement’s points might have been compared to points that were collected during a different time with other crane- and straddle carrier operators. After the extreme outliers were excluded a z-test was used if the data contained more than 30 data points. To perform a statistical test on the measurements they were subtracted with each other. The data for the second half hour was subtracted 32 3. Methodology with the data for the first half hour for both tests. This gave one series of data for each test. The statistical test was then performed on these two subtracted data series, the one when comparing the extra information and the one comparing the difference between the first and second hour. When performing the statistical test, the z-value is calculated. As stated in equation 2.2, there are three things that are needed to calculate the z-value. Those three things are: the sample size (n), the mean value (x) and the standard deviation (σ). All three values were calculated in Matlab which can be seen on row 77 to 79 in appendix C. The mean value was calculated by using the mean-function on the collected data, the standard deviation by using the standard deviation-function and the sample size was calculated with the length-function. The z-test was used to analyze if there was a statistical significance between the first and second half hour for both tests. Before performing the z-test a hypothesis was stated with a null hypothesis (H0) and alternative hypothesis (HA). The objective was to see if the loading time for each container decreased between the first and second half hour. Therefore, the null hypothesis stated that the average time for a loading cycle was the same or greater during the second half hour. The alternative hypothesis stated that the average time for a loading cycle was less during the second half hour. The stated hypothesis can be seen below were y1 was the average time for a loading cycle during the first half hour and y2 for the second half hour. Hypothesis: H0 : y2 − y1 ≥ 0 HA : y2 − y1 < 0 This hypothesis is called a lower-tailed test which is described in section 2.6.2. If the z-value was lower than -1.645, which is the z-value at a 5% significance level for a lower-tailed test (Sullivan, 2017), the null hypothesis was rejected and the alternative hypothesis accepted. This implied that there was a difference between the two half hours and that the STS crane performed better during the second half hour, hence an improvement in the case terminal. If the z-value was greater than -1.645 the null hypothesis was accepted, hence no improvement in the case terminal. 3.5.1 Example of the calculation procedure To show the statistical test procedure, an example with 30 data point for each half hour was made. The data for the example is given in appendix B, it was collected from StarDriver and the duration for each motion was randomized between 40 and 60 seconds. The hypotheses used for this example test was the same as the one used in this thesis. The example data was put into one Excel document and split up into two different Excel-sheets. The data for the first half hours was put into the first sheet and the data for the second half hours was put into the second sheet. The values in the excel document worked as an input in the Matlab program. The data 33 3. Methodology was divided by the median for the first half hour to mirror the procedure for the tests in this thesis. Matlab calculates the boxplot which is visualized in 3.3. Figure 3.3: Boxplot over the example data from appendix B. The extreme values for the upper-limit was calculated in Matlab and for the first half hour the extreme upper-limit was 1,49 and for the second half hour it was 1,44. The max values were also calculated in Matlab and they were 1,14 and 1,15 respectively. Since none of the max values were greater than the upper-limit for the extreme outliers no data points were excluded. The difference between these two tests are calculated by subtracting the second half hour with the first half hour which is performed on row 75 in the code in appendix C. The number of measurements (n), mean value (x) and standard deviation (σ) were calculated in Matlab and they are given in table 3.5. Table 3.5: Results from the example provided in appendix B. Mean value x1 -0,018 Standard deviation σ1 0,138 Number of measurements n1 30 Since the number of measurements (n) are 30 or above, a z-test could be performed. With the numbers in table 3.5 the z-value was calculated in Matlab. The equation used for calculating the z-value is given in equation 2.2 and used in equation 3.1. Z = √ n ∗ (x) σ = √ 30 ∗ (−0, 018) 0, 138 = −0, 727 (3.1) 34 3. Methodology Since the z-value was greater than -1.645, which was the z-value at a 5% significance level for a lower-tailed test (Sullivan, 2017), the null hypothesis could not be rejected. Therefore, it was not statistically significant at a 5% significance level that the mean for the second half hour was lower than the mean for the first half hour. The p-value was also calculated in the Matlab code and it was equal to 0,766. This gave that there was a 76,6% chance that mean loading cycle in the second half hour was lower than the mean loading cycle in the first half hour. Since the p-value was lower than 95% it was not significant at a 5% significance level. 3.6 Reliability and validity It is the credibility and feasibility of a project that determines if others accept it (Bryman, 2012). One recommended technique to ensure feasibility and credibility is triangulation, triangulation requires that more than one source of data is used during the study. The term refers to a method that uses various source of data, theoretical perspectives and observations. In this thesis, multiple sources of data were used to increase the trustworthiness and validity. According to Bell (2011) interviews usually takes quite some time and during a project there might not always be room for more than a few interviews. Therefore, the risk of bias could be large. The interviews in this thesis were always done anonymously to minimize the risk that the interviewed person would be restrained. There was also more than one interview for each competence to reduce the impact of one individual interviewee. One of the most obvious main elements for the reliability of measurements is that the same answer should be obtained even if measurements are carried out during two different occasions (Bryman, 2012). In this project a statistical hypothesis test was used to ensure the reliability of the measurements. The z-test gave the probability that the same result would be obtained if measurements would be done on the exact same thing again. If the probability is high, the reliability of the measurement should be considered large. Another thing is the validity of the measurements. It is important that the test really measure the stated concept (Bryman, 2012). To guarantee the validity of the measurements the tests were carried out in two versions. The second version was made to see so that nothing else than the shared information impacted the result for the first test. 3.7 Earlier experience from seaports As already mentioned, one of the authors previously worked in the case terminal. This meant that the termin