Pickup and Delivery Efficiency
A study regarding the allocation of tasks within the pickup and delivery process
Page | II
Name: R. van der Ploeg
Student number: s1802739
Master program: Business Administration:
Operations & Supply Chains
Year: 2009/2010
Company: TNT B.V. Duiven
University: University of Groningen
Supervisor TNT Express: Mr. T. Reeves
First examiner: dr. G.A. Welker
Second examiner: dr. N.D. van Foreest
Page | III
Preface
This thesis represents the concluding part of the Business Administration Operations and Supply Chains master degree program at the University of Groningen. TNT offered me the opportunity to apply theoretical knowledge in practice. By performing this research my knowledge and understanding of operations and supply chains has increased.
I am grateful for the persons who made it possible to get in touch with TNT. In addition, I would like to thank everybody that were directly involved in my research. The execution of this study was not possible without the willing cooperation of the employees working at Duiven and the several depots I have visited. My sincere appreciation for this. I am also grateful for my supervisor Terry Reeves. I would like to thank him for his help and valuable advises during the study. Special thanks goes to the supervisor of the University of Groningen; Gera Welker. During the execution of the study she provided valuable feedback and made it possible to complete the thesis as planned.
Finally, I owe a huge thanks to my family for their untiring support and for making it possible to start and complete my masters degree.
Dalfsen, 17 August 2010
Page | IV
Abstract
Page | V
Index
Preface ...III
1. Introduction ...1
2. Description of the organization ...2
2.1. TNT B.V. ...2 2.2. TNTs distribution network ...2 2.3. PUD-process ...4 3. Research design ...9 3.1. Research motivation ...9 3.2. Research scope ...9 3.3. Research model ... 11 3.4. Research objective ... 12 3.5. Research methodology ... 14 3.6. Research outcome ... 15 4. Results ... 17 4.1. Loading methods ... 17 4.2. Examined depots ... 18 4.3. Overview of findings ... 28
4.4. PUD operational costs ... 31
5. Findings... 33
5.1. Factors influencing the allocation of tasks ... 33
5.2. Design of decision model ... 35
5.3. Discussion ... 36
5.4. Further research ... 37
6. Bibliography ... 38
Appendix A: Linehaul arrival times ... 40
Page | 1
1.
Introduction
With the influence of the economic crisis and the ongoing competition on the global express market, the focus on cost reduction is higher than ever before. Cost optimisation is vital because prices must be competitive. Therefore TNT Express is trying to optimize the execution of all processes performed within TNT Express. This study will emphasize on one of these many processes; the pickup and delivery process.
To realize the optimization of the pickup and delivery process (further called PUD process), several depots have been visited. During these visits data has been gathered to gain more insight in the way of working at the different depots. The outcome of this study would be a decision model which is applicable within European depots. This paper reports on the execution of this study.
Page | 2
2.
Description of the organization
This section will introduce the company in which this study has been conducted. It will start with a short general description of TNT B.V. followed by elaborating TNT’s distribution network. The section closes with the short introduction of the involved PUD process.
2.1.
TNT B.V.
TNT N.V. has two divisions, Express and Post, and is part of the global transportation and distribution industry. TNT Post is mainly focusing on the post market in the Netherlands while TNT Express is among the four largest Express organizations worldwide. The ‘TNT Express’ division offers dedicated services options to its customer and where applicable will tailor solutions to meet specific customer requirements. These requirements are divided among the five generic performance objectives; speed, cost, dependability, flexibility and quality (Slack & Lewis, 2008). In 2009 TNT Express had a market share of 18% with DHL (16%) and UPS (9%) as main competitors and a total net sales of more than 10 billion euros (TNT annual report, 2009). TNT’s mission is “to exceed its customers expectations in the transfer of their goods and documents around the world. TNT delivers value to its clients by providing the most reliable and efficient solutions through delivery networks.“ (TNT annual report, 2009).
This research is executed within the Business Unit ‘TNT Express Global Networks & Operations’ which is part of TNT Express Road Network B.V. (TNT ERN). TNT ERN is responsible for organizing all the international road connections of TNT Express in Europe. Currently the road network has connections with 29 European countries and more than 750 departures a day. The ‘International Road Hub’ in Duiven, where TNT ERN is based, has 162 loading/unloading docks and an automatic sorter which is sorting on average 5.000 tons a day. The hub employees approximately 500, temporary as well as full-time, employees working in shifts (TNT Express). Within TNT ERN several studies are currently being executed. One of them is the ‘Global Optimization program’ or ‘GO’. The goal of this program is to optimize the operations and supply chains of TNT Express worldwide (TNT annual report, 2009). To get familiar with the operations and supply chains of TNT Express, the distribution network will be discussed in the next section.
2.2.
TNTs distribution network
Page | 3 while the process between the customers and the depots is called ‘Pick Up and Delivery’, or short ‘PUD’, which will be elaborated in next section.
Figure 1: A typical hub and spoke network.
Page | 4
Figure 2: Overview of TNT Express Road Networks hubs and depots connected by linehauls.
All depots and hubs are linked together by the linehauls forming TNT’s Express Road Network. In order to guarantee fast distribution among this network, a high precision linehaul-schedule is used for this road network. This schedule is very complex because a lot of external factors are influencing the schedule; for example the law. A driver can drive his truck for a limited number of hours per day. Linehauls covering long distances are therefore equipped with two drivers to avoid stagnation of the linehaul. If a driver needs the compulsory break, he will be replaced by the other driver. Constructing an air schedule is even more complex, not to talk about a combination of the road and air network. In short, it is nearly impossible to adjust just one linehaul without affecting another linehaul, thus this should be avoided as a recommendation of this study.
2.3.
PUD-process
Page | 5 consignments up at the assigned address. Therefore the process is called the ‘pickup and delivery process’. For the sake of readability a pickup or delivery is abbreviated to a stop. At the beginning of a working day the PUD-vehicle will be loaded with the consignments assigned to a certain round. At the end of a working day the PUD-vehicles unload the consignments which have been picked up during a round. During the evening and sometimes the night, sorting takes place to determine to which depot the consignments should be distribute to. The PUD fleet performing the rounds of TNT Express consist of 33.000 PUD vehicles of various types. The lifecycle of a consignment is illustrated in Figure 3.
Figure 3: The lifecycle of a consignment.
Page | 6 the service area of the current depot the consignments stay at the depot in order to be delivered the next PUD-round. If not, the consignments are supplied into the ‘linehaul’ which is the end of the PUD-process. This study will focus on the orange marked area as illustrated in Figure 4.
Figure 4: Schematic overview of the PUD-process.
As Figure 4 illustrates, the processes conducted at the depot consists of processing and (un)loading consignments. These processes are conducted at every depot using the same outline. The current and desired state are illustrated in Figure 5. Each part of the processes conducted at the depot will be discussed in the next sections.
2.3.1.
AM-process
Page | 7 When the linehauls arrive the actual processing will begin. Obviously the linehauls must be unloaded first. After that, the consignments are sorted and transported to the correct loading bay by using cages or pallets. This is called the ‘primary sort’; the sort to the appropriate loading bay. Thereafter the consignments are loaded in a PUD vehicle. The driver determines the route he will drive and thus the location of the consignments in the vehicle so the appropriate consignments can be found quickly during the round. This is called the ‘secondary sort’. When all consignments are processed and loaded into the PUD-vehicle, the driver will get his run sheet (the document containing detailed information of the consignments allocated and loaded in his round/vehicle) and he can start his round.
Figure 5; A detailed overview of the different steps in the PUD process performed by a driver.
2.3.2.
Stem time AM/PM and break
The stem time is the time a driver needs to travel to and from the PUD area. It depends on the round and other external influences how long this stem time will be. This time is therefore not adjustable. Needless to say a break is necessary as well as compulsory and cannot be affected either. Both the stem times and break time are for that reason out of this projects scope.
2.3.3.
Productive time / Road time
The productive time is the time the actual stops are performed. In order to maximize the number of stops this time is required to be as long as possible. The longer the productive time the more stops can be made, as indicated by the desired state. The productive time will be abbreviated to road time.
Page | 8 time consumption of the AM and PM process (indicated red) which provide more productive time and thus servicing more micro-zones (indicated yellow). TNT conducts therefore studies to contribute to the transition from the current state to the desired state.
2.3.4.
PM-process
When the day ends and the PUD-rounds are finished, the PUD vehicles return to the depot. The PM-process is similar to the AM-PM-process. The consignments picked up during the round or which were undeliverable are unloaded. After unloading, a sort will be made in order to determine which consignment can stay at the depot or which will have to be transported to another depot. In the last case, linehauls are used again to provide transportation towards hubs. Because of only a small number of hubs involved instead of many different PUD-rounds, the sort does not have to be as thorough as the AM-process which makes the PM-process simpler.
Page | 9
3.
Research design
This section describes the research design. It will start with the research motivation why this study has been conducted, followed by the scope of the study. After introducing the research model, the research objective and methodology will be elaborated. The section closes with the desired research outcome.
3.1.
Research motivation
As mentioned in the introduction, the focus on cost reduction is higher than ever before. Cost optimisation is vital because of the ongoing competition. Therefore TNT Express is trying to improve the performance of all his processes. This study will examine the PUD process. The reason to choose for the PUD process is because it is responsible for a reasonable part of the total costs of TNT Express, 40% of total cost globally (TNT annual report, 2009) A slight performance increase is therefore assumed to have a big positive influence on lowering the global operational costs. To be able to have a thorough research, TNT Express wanted the research to be conducted from an academic perspective. Therefore this study has been set up.
3.2.
Research scope
This research will focus on the PUD-process and specifically on the activities which have to be conducted at the depot concerning PUD. The scope is also indicated in Figure 4 by the squared marked area ‘Depot’ at the left hand side.
Page | 10
Figure 6: The time constraints and connectivity concerning a PUD driver.
But why does the road time needs to be increased? The answer to this is to increase the number of stops a driver can make. To be able to increase the performance of PUD, the number of stops of a PUD round needs to be higher (Schönsleben, 2007). There are more factors influencing the number of stops, but with respect to the available time, this study will only consider the road time. The hypothesis is that with more stops per round less vehicles will be needed and thus the operational costs of the PUD process will be reduced. An additional possible advantage is the influence on the environment. With less vehicles on the road the CO2 emission is likely to be reduced. However, this study will only concentrate on lengthening the road time and will leave other factors influencing the performance out of scope.
Page | 11 deliveries/pickups’. Each depot initially allocates dedicated performers to the tasks themselves. However, there are depots using a mix of performers;
A. Vehicles that are loaded by the PUD driver themselves B. Vehicles that are loaded by the warehouse operatives.
C. Vehicles that are (un)loaded by either a warehouse operatives (primary sort) or the PUD driver (secondary sort).
The scope of this study will be to compare and analyse depots, each using a different method, referring to A, B and C, in order to deliver a representative model of the PUD-process for TNT Express Operations Europe. For that reason the model has to be generizable to the linehaul network of TNT Express operations Europe.
All three methods (A, B and C) are currently performed in different depots. During this study depots using these methods will be visited. The hypothesis, based on experience with and performances of the different methods, is that there are major benefits in PUD efficiencies, performance increase and significant cost savings for TNT by adopting method B as a common process where ever the conditions exist to allow this. In other words, TNT is expecting method B is ‘best practice’. Additionally, local differences are influencing the PUD process. In practice, these differences could be the arrival schedule of the linehauls or the type of linehaul. Section 4.1 will elaborate more in detail on the different methods of loading.
3.3.
Research model
The goal of the research model is to render a visual overview of the independent and dependent variables. In this case the dependent variable is the road time of the PUD-process. This variable is directly influenced by the allocation of the loading methods. This allocation is influenced by local differences like the arrival of linehauls, the type of the linehaul (truck or airplane) and other factors at the depot. In turn, the road time determines the performance of the PUD round; the longer the road time, the more stops can be made.
Page | 12 The variable which directly influence the allocation of loading methods are the local differences. It differs per depot what the road time of the PUD-round will be because of, for example, the arrival of linehauls and the type, airplane or truck, of the linehaul. But also other local factors can have influence on the allocation of the loading method.
Next to that, the allocation of loading method and the additional road time has influence on the operational costs which in turn influence the performance of the PUD round. This research model and the relations will be elaborated more in detail in section 3.4 in order to elaborate, explain and to justify the research questions.
Figure 7: The research model
3.4.
Research objective
Based on the research motivation and the research model a research objective has been composed. The research objective explains why this project is being executed. What does TNT want to achieve with conducting this project? With conducting this study TNT tries to improve the performance of the PUD process
Research objective:
Improve the performance of the PUD process by lengthening the road time.
Page | 13 thing’. Sink and his colleagues (1984) on the other hand defined seven dimensions in order to capture their conception of what performance means. These seven dimensions are effectiveness, efficiency, quality, productivity, quality of work life, innovation and profitability/budgetability. Following these papers, you may say productivity is part of performance.
Since this study is a research-study the research objective will be met by using a research question. With support of the research model and research objective, the following research question has been formulated:
Research question:
What is the optimal allocation of tasks within the PUD-process in order to increase the ‘road time’ of the PUD-process and what is the accompanying influence on the total operational costs?
To be able to answer the research question with taken into account the research model, sub questions have been formulated. Sub questions serve as a support for answers to the main research question and ensures a thorough study.
The first sub question is based on the current situation and used standard way of working methods. Taking into account the several involved depots, the effect of the current way of working of the PUD-process on the road-time needs to be investigated. However, the majority of literature is concentrating on the routing problem with simultaneous pickup and deliveries (Zachariadis, Tarantilis, Kiranoudis, 2009), (Bianchessi, Righini, 2007) and (Wassan, 2006). This is not applicable within this study, since it is no routing problem, and demonstrates the presence of a gap in the current literature. This study will therefore try to help filling this gap just like Slack, Chambers and Johnston (2007) did. They are stating that the performer of the task (or operation) in the supply chain is very important. Therefore it is assumable this occurs at the involved depots as well. To find out if this is applicable in the involved depots the first sub question has been formulated as follows:
Sub question 1:
What is the effect of the current way of working in the involved depots on the road time of the PUD-process?
Page | 14 firm needs to examine the interaction between its human resource strategy and its logistics strategy. Not addressing this strategic fit may lead to reduced optimization in the functioning of the supply chain (McAfee, Glassman, Honeycut, 2002). This aspect is less usable in this study since the influence of human research is out of scope. This study will investigate the more practical local differences which are determined more by the environment and location of the depot. A depot on a island is assumed to be supplied by a type of linehaul compared to an depot on mainland; airplanes rather than trucks. Both types and the distance to a hub influence the arrival of linehauls. Therefore the second sub question will provide more insight into the local differences variable.
Sub question 2:
What are the local differences which have reasonable influence on the PUD process?
The next step of this study is focusing on the current operational costs of the current way of working at the several involved depots and what the influence on the operational costs of the PUD process. (Simchi-Levi, Kaminsky, Simchi-Levi, 2008). Schönsleben (2007) has dedicated a chapter of his book to costs, cost elements and cost structures. He states that many organisations use standard cost accounting system because of difficulties associated with actual costing systems. This can be seen within TNT as well. TNT uses a standard cost accounting system which will be elaborated more in section 4.4. The third sub question is based on the current operational costs of the current way of working at the several involved depots is stated as follows:
Sub question 3:
What is the influence of a changing work process and road time on the PUD operational costs?
If all sub question have been answered, the main research question can and will be answered followed by the completion of the research objective.
3.5.
Research methodology
Page | 15 As mentioned before there are different (un)loading methods used along the depots. Next to that, other local factors are influencing the performance of the PUD processes. Therefore in the ideal situation data is needed from every depot, which is unrealistic. Therefore data will be collected from a sample of depots. A total of four depots will be involved. Each depot has his own characteristics regarding the local differences and allocation of loading methods and will provide useful input for this study. The depots involved are:
United Kingdom - Liverpool
Ireland - Belfast
Belgium - Kortrijk
France - Paris
Each depot will be visited for a week (4 to 5 days). From each depot data will be gathered and analyzed to provide valuable input for the research outcome. Necessary data will be gathered by conducting interviews with the operation manager and general depot manager of the depot. During these interviews the focus lied at facts and figures to get the most reliable data. Next interviews, observations will be made at the shop floor during different shifts. If applicable, time studies will be conducted in order to get a decent impression of how the different methods are performing at each depot. However, because of the limited visiting time, the time studies will be based on samples and averages. Where possible, data will be retrieved from existing databases managed by TNT. Using this method, time consumption of several process is examined. During the execution of this study theory will be intensively consulted as well. This theory will be retrieved from books and peer reviewed journals using electronic databases like ‘Business Source Premier’ and ‘Purple Search’. Using this approach a comparison and analysis of processes at each depot will be made.
3.6.
Research outcome
Page | 16 harmonized throught the chain, which in turn helps to improve the supply chain performance. Therefore the desired research outcome needs to be a widely applicable decision model which ensures harmony between depots. Each depot with similar charasteristics will perform their process similar as well.
There are different types of decision models. Possible decision models are decision matrix, pareto chart, flow chart, cause and effect diagram and a decision tree (Fitzsimmons, Fitzsimmons 2008). The choice has been made to construct a decision tree. According to Abellán and Masegosa (2010), a decision tree is a simple structure where each node represents a variable and each branch represents one of the states of this variable. The ‘leafs’ of the tree specifies the expected value of the variable depending on the information contained in the data set. This is precisely what TNT wants; the appropriate loading method based on the charasteristics of the involeved depot.
Page | 17
4.
Results
This section starts with providing more insight in the different loading methods with information provided by the depot visits. The depot visits themselves will be elaborated in section 4.2 followed by an analysis to determine which factors, emerged from the site visits, are influencing the allocation of tasks in the PUD process. At the end of this section the operational costs will be discussed.
4.1.
Loading methods
As mentioned earlier there are distinctively three methods of loading a PUD-vehicle. This section will elaborate more on these three types of allocation. As mentioned in section 2.3.1, the primary and secondary sort can be performed by the PUD driver as well as the warehouse operative.
4.1.1.
Full warehouse operative
The first method is with the first and secondary sort performed by the warehouse operative completely. The warehouse operatives are not using any cages because there is no need to buffer consignments at the loading bays. They are directly loaded into the vehicle. When the PUD driver arrives at the depot, he only has to retrieve the run sheet and start his round nearly immediately. However, he has to spend some time in studying the run sheet to determine his route for the day. This method is used at the Liverpool depot.
4.1.2.
Full PUD driver
This method is essentially the same as the method whereby the loading is done by the warehouse operatives. All handlings stay the same but they are performed by the driver instead of the warehouse operatives. When the driver start his shift the consignments are not buffered at the loading bay yet. All consignments are still inside the cages which were unloaded of the linehauls. This method is used at the Belfast Airport depot.
4.1.3.
Combination warehouse operative and PUD driver
Page | 18
4.2.
Examined depots
Depots located in four different cities have been visited; Liverpool, Belfast, Kortrijk and Paris. The depots are operating comparable to the cross-docking method (Simchi-Levi, Kaminsky, Simchi-Levi, 2008). The depots do not hold inventory (consignments) and the only thing they do is rearranging the content of the inbound linehauls into the PUD rounds. All depots will be discussed in this chapter in order to answer the first sub question: What is the effect of the current way of working in the depots on the road time of the PUD-process?. Only the AM and PM process will be elaborated on, since this is the research scope as stated in section 3.2. The methodology of how the time study is conducted is elaborate din section 3.5.
4.2.1.
Liverpool depot
The Liverpool depot is located at the John Lennon Airport, however the vast majority is distributed by truck. The airplane is rarely used as a regular linehaul. Figure 8 provides the layout of the depot. The linehauls are docked and unloaded at the top left. The PUD vehicles are docked at the loading bays if they are not performing the PUD round. Mainly small trucks are used for performing the rounds and handle both pallets and small consignments. The average number of stops per PUD round is 42.
Figure 8: Layout of the Liverpool depot.
AM-process
Page | 19 so it can be placed inside the trailer avoiding unnecessary transportation and carrying. The conveyer belt splits into other conveyer belts leading to the loading bays with the corresponding PUD-round. A scanning device at the beginning of the belt, automatically allocates the consignments to the correct conveyer belt. In turn, a warehouse operative picks up the consignment from the belt and loads it directly, after scanning it, into the vehicle. One warehouse operative is loading about five to six vehicles and in total 30 PUD rounds are carried out. At 9am the processing at the depot is finished and the drivers arrive at the depot to start their working day. They collect their run sheets and they can start with their round immediately. The average time consumed by the AM process , the time between the arrival time and the departure time of the driver about 20 minutes.
PM-process
At the end of the day the drivers park their vehicle at the loading bay again, de-brief and go home. Their work is done. At 6pm, the consignments are unloaded and sorted by the warehouse operatives. Again the conveyer belt is used to transport the consignments directly into the linehaul or towards a buffer when the delivery address is serviced by the Liverpool depot the following day. The average time between the return time and the finish time of the driver is approximately 15 minutes.
4.2.2.
Belfast depots
Belfast has two separate depots. One is located at the airport and one is located just outside Belfast city. The airport depot is mainly used for international consignments while the Belfast depot is used mainly for domestic consignments. For the sake of readability ‘airport’ refers to the airport depot and ‘depot’ is referring to the Belfast depot. Both depots are independently and act as starting point for PUD-rounds. The average number of stops per PUD round is 42.
Page | 20
Figure 9: 'Igloos' being unloaded from an airplane.
4.2.3.
Belfast City depot
The Belfast depot is approximately 25 years old and there has not been made an investment for over 10 years. The depot has 15 loading bays with manual operating doors of which 12 are used to feed approximately 40 PUD-rounds. Both trucks and vans are used to perform the rounds and carry both pallets as well as small consignments. Because of limited number of loading bays a lot of movement of vehicles is needed at the area outside the depot. A lay out of the depot is shown in Figure 10.
Page | 21 AM-process
Normally, the inbound consignments by truck arrive in the morning at approximately 4 am. Close to this time the day shift starts working. While the inbound of igloos are allocated to either one of the Belfast depots, the linehaul truck is mixed with consignments for both depots. Therefore, the sort has to be conducted first to distinct the consignment from both depots. After that, consignments are loaded into a truck which will transport them to the airport. The igloos arrive at the depot at around 8 am.
Because the number of loading bays do not correspond with the number of PUD rounds, the vehicles are double or even triple parked at the loading bays. Therefore a lot of vehicle movement exist at the depot site and buffering consignments for all rounds on the shop floor is needed.
Next to that, it is a small depot and the space is limited. A daily struggle is taking place to utilize the available space to its maximum. As mentioned before the depot is small and at peak hours the shop floor is filled with consignments which makes manoeuvring with a lift truck difficult without making damages. The drivers start working between 6 and 7 am. The average time consumed by the AM process , the time between the arrival time and the departure time of the driver, is about 2 hours.
PM-process
The PUD vehicles arrive back at the depot in the afternoon. The driver unloads his cargo and as soon as possible a first sort is taking place by the warehouse operatives in order to meet the departure time of the airplane which is planned to be 19.30. No cages are used, because of the size of the depot. The consignments are loaded into igloos which in turn are loaded into a truck which transfers them to the airport. The consignments with a destination other than the airport are loaded into a truck which will transfer them to Dublin. The average time between the return time and the finish time of the driver is 15 minutes.
4.2.4.
Belfast Airport depot
Page | 22
Figure 11: Lay-out Belfast Airport depot.
AM-process
Every morning an airplane arrives, with a stop in East-Midlands, from Liége. The airplane is planned to arrive at about 7 am. When the plane has arrived the unloading process starts and the igloos are either transported to the depot or stay at the airport. Each igloo contains only consignments which are allocated to either the depot or the airport.
Page | 23 date. The average time consumed by a driver during the AM process , the time between the arrival time and the departure time of the driver, is about 2 hours.
PM-process
The airport does not process outbound consignments. As mentioned before the PUD vehicles unload at the depot when they finished their round. They only receive igloos from the depot and load them in the airplane.
4.2.5.
Kortrijk depot
The Kortrijk depot is a newly build depot. It is build ‘for the future’ so at this time the shop floor is more than sufficient. Cages or pallets are used for transportation on the shop floor between vehicles and conveyer belts. Unlike Belfast and Liverpool, Kortrijk has two separated types of PUD rounds; ‘truck’ and a ‘van’ rounds. The trucks are handling big and heavy consignments, often pallets. Vans are handling the smaller and less heavier consignments. Obviously the trucks are handling smaller consignments as well when a pallet has to be delivered to the same address. The layout of the depot can be found in Figure 12.
The average number of stops for a van round is 55 per round, while the trucks have an average of 20 stops per round. Taking both kind of vehicles into account results in an average of 34 stops per round.
Figure 12: Lay out Kortrijk depot, with on the right the processing of pallets and on the left hand side the processing of small consignments.
AM-process
Page | 24 the last linehaul has been processed. However, it can happen the linehauls holds consignments for particular PUD rounds only; it is unknown to the warehouse operatives what this last linehaul consists of. This results in unnecessary waiting time for drivers not involved in the freight of the last linehaul.
The warehouse operatives start their shift at 5am as well. The warehouse operatives sort the inbound consignments into cages allocated to the PUD rounds. When the drivers are starting at 8am they unload the cage en load them into their vehicle. Some drivers first sort the consignments on the shop floor before loading them, whilst others load directly from the cage into the vehicle. It depends on the drivers experience and preferences which method is used. The average time consumed by the driver during the PM process , the time between the arrival time and the departure time of the driver, is 1 hour and 10 minutes.
PM-process
After conducting the PUD round the drivers return back to the depot. They unload their vehicle and in the meanwhile they make a first sort; domestic or international cargo. The reason for this is the higher priority of international compared to domestic consignments. The chance of consignments exceeding their promised delivery date is reduced when both types are distinguished early. The average time between the return time and the finish time of the driver is 22 minutes.
4.2.6.
Paris depots
Because of a recent take-over, TNT Express France distinguishes two types of consignments; domestic and international. Both are treated completely independently with both having their own depots and PUD-rounds. To gain complete knowledge, both kind of depots have been visited. Bonneuil-en-France for domestic consignments and Créteil for international consignments.
Page | 25
4.2.7.
Paris Bonneuil-en-France depot
Bonneuil-en-France is a national depot. Which means it will (and can) only process domestic consignments. The depot at Bonneuil-en-France is the starting point of 65 PUD rounds and employs approximately 100 employees. A vehicle is loaded with maximum 50 to 60 consignments. The average number of stops is 39. The layout of the depot can be found in Figure 13.
Figure 13: Layout of Bonneuil depot, with a conveyer belt for the AM process and one for the PM process
AM-process
The AM-process starts at 1 am in the morning until approximately 8 am. During the AM-process the inbound of consignments consists of about 15 linehauls with the first arriving at 23am. Till 5.30am 7 linehauls arrive with a steady rate. Contradictorily, between 5.30am and 7 am 7 linehauls are planned to arrive as well. In other words, during these 1,5 hours almost 50% of the inbound is arriving, which causes rush hour at the shop floor.
The linehauls are unloaded and the consignments are sort into cages corresponding to the rounds. This primary sort is always done by the warehouse operative. The drivers and their vehicle arrive between 5 and 6 am, parking their vehicle at the loading bay followed by the actual loading of the vehicle. 90% of the drivers are not loading directly into the vehicle. They first make a sort on the shop floor before loading them into the vehicle. This secondary sort is completely performed by the drivers and takes approximately 1,5 hours.
Page | 26 In the afternoon the PUD drivers drive back to the depot. They unload their cargo into cages, which will be sorted by warehouse operatives. A sort is made on department base1. Each depot serves certain departments, thus if you know which department the consignments are allocated to, you know which depot is the destination. Linehauls are loaded to transport consignments to the destined depots. The average time between the return time and the finish time of the driver is 30 minutes.
4.2.8.
Paris Créteil depot
The Créteil depot is only used for international consignments. The national consignments will be processed using other depots, like Bonneuil. Créteil is the starting point of approximately 100 rounds, it differs daily. The average number of stops lies at 27, which is quite low which is caused by the very busy service area of the depot. The layout of the depot can be found in Figure 14.
Figure 14: Lay out Créteil depot, with vehicles driving on the shop floor because of too few loading bays.
AM-process
At Créteil the shift who is performing the AM-process starts at 3 am. The first linehaul during this shift will arrive at about 4 am. The hour of slack is used to process consignments which stayed at the depot the previous day. The linehauls are arriving on a steady rate. However, most of the time extra linehauls are planned. So in practice the arrival of linehauls are shifting near the end of the AM-process.
1
Page | 27 However, agreements have been made with major Parisian customers that if they supply the consignments to TNT before midday (12pm) the consignments will be delivered the same day to addresses in Paris. To achieve this, the drivers are having their lunch break at a agreed place together. During this break the consignments are redistributed among the PUD vehicles and after the break the drivers will deliver them. Using this method the consignments will reach their destination the same day. Obviously, drivers performing their rounds close to the depot will have their lunch break at the depot and they can therefore load the extra consignments directly at the depot. When a driver knows he is returning back to the depot, he can leave some consignments at the depot to avoid an overloaded truck. The average time between the start time of the driver and the departure of the driver is 1:24 hours.
PM-process
Page | 28
4.3.
Overview of findings
As mentioned in section 4.1 distinctively three different methods of loading can be used. Derived from the site visits, these three types of allocations of tasks are all having their advantages and disadvantages. A comparison can be found in Table 1.
Full warehouse operative
Combination
Full driver
Advantages:
• TNT in control of process
• Quick and smooth process
• Little use of buffers
• Shortest in-house time of consignments • Exchange of consignments between rounds • Locating freight • Balanced knowledge of process
• Driver is familiar with consignments
• Easy exchange of consignments between rounds
Disadvantages:
• Drivers is not familiar with consignments more time needed during round
• Early linehaul arrival required
• Vehicle and loading bay availability • Usage of buffer • Multiple handlings • Much transportation needed • Subcontractors in control of process
• Longest in-house time of consignment
Table 1: Comparison of the three loading methods.
Page | 29 his round. Drivers may exchange these consignments avoiding unnecessary driving time. The disadvantage is that the drivers, often subcontractors, are in control of the process at the shop floor which are less familiar with the processes. The combination of warehouse operative and driver is a combination of the (dis)advantages of both other methods.
To be able to provide a clear overview of the similarities and differences of the involved depots and loading methods, Table 2 has been constructed. The criteria mentioned in Table 2 are al derived from the site visits and are mainly concerning the arrival of the linehauls which appeared to be very important in the allocation of tasks.
Liverpool
Belfast
2Kortrijk
Paris
Airport Depot Bonneuil Créteil
Process time 12am-9am
(9hrs) 7am-9am (2hrs) 5am-9am (4hrs) 5am-9am (4hrs) 1am-9am (8hrs) 4am-9am (5hrs) Peak in Arrival? (% arrivals in last 1/3 of process time) No (38%) Yes (100%) Yes (75%) No (45%) Yes (71%) Yes (50%)
Arrival last linehaul 6:00 7:20 7:30 8:30 7:00 7:30
Primary sort by WO Driver WO WO WO WO
Secondary sort by WO Driver Driver Driver Driver Driver
Used loading technique (direct / via shop floor)
Direct Direct Both Both Shop
floor
Shop floor
Vehicles available during AM process
Yes Yes Yes/No No No No
Time consumption driver AM process 0:20 2:00 2:00 1:10 1:30 1:24 Time consumption driver PM process 0:15 n/a 0:20 0:22 0:30 0:25
Table 2: Depots compared using several factors.
The process time is the time the process is being performed; from the start of the shift till the depart of the PUD vehicles. The peak in arrival is estimated to occur if during the last 1/3 of the process
Page | 31
4.4.
PUD operational costs
As stated in section 3.4 the third sub question is: What are the current PUD operational costs taking into account the used way of working? This section will elaborate this. Since the main objective of this study is not concerning the costs aspect predominantly, only the direct costs of the PUD rounds will be used in this section.
Operational costs consists of two types of costs; fixed and variable costs. (Schönsleben, 2007) Fixed costs are the costs that remain the same regardless of the level of production (in this case number of rounds) and/or sales. The fixed direct costs of a PUD round consist of the rental or lease, repair and maintenance and insurance of the vehicle. The PUD fleet consists the bigger part of vans and trucks which are for the majority owned by sub contractors. Both types of vehicles have different fixed costs. The fixed costs per van is €XX and per truck €XX. Next to the fixed direct costs of the vehicles, there are also indirect costs concerning fleet management, dispatch/planning and other PUD transport costs like communication and scanners. These costs are calculated to be €XX. These fixed costs are calculated with the assumption there are no different costs structures per sub contractor and the extra costs for overtime are left out as well. All external or indirect costs are left out of the calculation.
Page | 32
Fixed costs Variable costs
Direct costs: Costs per hour:
− Rental/lease
− Repair/maintenance − Insurance
.
− Gross costs per driver − Social security contributions − Holiday allowance
− Bonus costs − Unavailability
= €XX = €XX
Indirect costs: Cost per km:
− fleet management − dispatch/planning − communication − scanners − fuel costs = €XX = €XX Van / €XX truck
Table 3: Overview of fixed and variable costs.
As stated in section 2.3, the PUD fleet consist of 33.000 vehicles globally. Combining this fact with all previous provided cost information, results in a saving of roughly €XX per day of PUD operational costs if the road time is lengthened, i.e. the AM/PM process shortened, by 10 minutes average.
Page | 33
5.
Findings
As stated in section 2.4, the objective of this study is to increase the performance of a PUD round by lengthen its road time. This resulted in the initial research question as formulated earlier: What is the optimal allocation of tasks within the process in order to increase the road time of the PUD-process and what is the accompanying influence on the total operational costs? As section 2.2 provided, the scope of this study lies at the AM and PM processes which can be performed by both the warehouse operative or the PUD driver. Obviously the driver is always needed to perform the actual pickup and delivery of consignments. However, he also may be involved in the primary and secondary sort process. Thus the question is; when and who should be allocated to the primary and secondary sort? Using the data gathered during the four depot visits, factors have been distinguished which are influencing this allocation of tasks. This research question will be answered in this section.
5.1.
Factors influencing the allocation of tasks
The PM-process is consuming a relatively low amount of time compared to the AM-process; respectively approximately 20 minutes and 90 minutes. Next to that, the PM-process consist of only unloading the vehicle and debrief whereas the AM-process consists of processing and loading the consignments. As can be seen in section 3 and 4, the variation of processes is bigger and more thorough in the AM process compared to the PM process. The AM process consumes more time as well. The best opportunity to lengthen the road time lies therefore at the AM-process.
As the results in previous section show, the AM-process is for the largest part dependable on the inbound of the linehauls. Three distinct factors were derived which have considerable influence on the allocation of tasks within the PUD process:
Page | 34 The second factor on the start time of the PUD round is the arrival time of the linehauls. The arrival time of the linehaul determines how much time is available to perform the primary and secondary sort. The earlier the linehaul arrives the earlier the work shift can start and the more time is available to process the consignments. The process time is the total time between the start of the night shift and the start of the PUD rounds. For example, Liverpool has a processing time of nearly the entire night, from midnight till 9am. Contradictory, Belfast Airport depot has a very short processing time because of the airplane which is landing at about 7 am. In this case the processing time is only 2 hours, from 7am to 9 am. More important is the time between the last linehaul and the start of the PUD round. The later the linehaul arrives, the less time is available to process the load of this linehaul. Under normal conditions unloading the linehaul and performing the primary sort takes approximately 1 to 1,5 hour. In other words, you need at least 1,5 hour prior to the start of the PUD round to process the consignments in the appropriate round under normal circumstances. When there is a peak in arrival and the processing time is short, like in the case of Belfast Airport, you need more workforce on a short notice. The easiest way to solve this, is to use the drivers. This aspect occurs, apart from Liverpool, in all depots concerning the secondary sort. Belfast Airport is even more extreme. At this depot both the primary as the secondary sort is performed by the driver. The reason for this is the process time. Liverpool has enough time to process the last linehaul, while Belfast has to deal with a late arrival of the airplane which contains more consignments compared to a truck.
The third factor influencing the allocation of tasks is the total time available to process consignments of the last incoming linehaul; the ‘window of opportunity’. This time window is the time available to process consignments of the last linehaul into the appropriate round. So if a linehaul is arriving more towards the start of the PUD rounds, the window of opportunity is getting smaller. If we look at the Liverpool depot, the last linehauls is arriving at about 6am. Thus there is almost 3 hours available before the PUD rounds will start; the window of opportunity is therefore 3 hours, which is the largest among all depots. At Liverpool all processing is therefore done by the warehouse operative. If we look at the Belfast Airport depot, the last linehaul is arriving around 7:30am; the window of opportunity is 1.5 hours, which is small taking into account the number of consignments arriving by airplane. All other depots have windows of opportunities which are between 3 and 1.5 hours.
Page | 35 the arrival peak, Liverpool has the most time to process the last linehaul, whereas Belfast Airport has the shortest. Kortrijk and Paris (Créteil and Bonneuil) are floating between these two extremes. Kortrijk is leaning towards the Liverpool method whereas both Paris depots lean more towards Belfast.
5.2.
Design of decision model
Using the influencing factors derived in previous section a decision model illustrated in Figure 15 has been constructed. If there is an arrival peak in linehauls, the processing time is short and the window of opportunity is small the processing (primary and secondary sort) should be performed by the driver completely. If there is an arrival peak in linehauls, but the processing time is long or the window of opportunity is large, the primary sort and secondary sort should be performed by respectively the warehouse operative and the driver. If there is no arrival peak in linehauls, but the processing time is short and the window of opportunity is small the primary and secondary short should be performed by respectively the warehouse operative and driver as well. If there is no arrival peak in linehauls, the processing time is long or the window of opportunity is large, the primary and secondary sort should be performed by the warehouse operative.
Figure 15: Decision model starting at the top left with the arrival of the linehaul.
Page | 36 The combination of the decision model and the (dis-)advantage overview in Table 1 provides the answer to the research question; What is the optimal allocation of tasks within the PUD-process in order to increase the ‘road time’ of the PUD-process and what is the accompanying influence on the total operational costs? The answer is that the warehouse operative is the best person to be allocated to process consignments. Observations show less use of buffers and a quicker and smoother flow of consignments at the shop floor resulting in a low in-house time of the consignments. This is also what Slack, Chambers and Johnston (2007) are stating. They say one of the most important approaches improving operational efficiency is known as ‘time compression’. This means speeding up the flow of materials (in this case consignments) down the chain and the flow of information back up the chain. With only one person performing both the primary and secondary sort, communication and exchange of information between the primary and secondary sort will become easier and faster. Speeding up the exchange of information by time compression in the supply chain results in an overall positive impact on profitability; costs will reduce and revenues will increase. A tool which can be used helping achieving this is the LEAN technique which increases efficiency, reduces unnecessary handlings (waste) and using empirical methods to decide what matters. (King, 2009).
As showed in section 4.4 every 10 minutes reduction of in-house time of the driver, and thus lengthening the road time, results in a global saving of roughly €XX each day. This is another reason to allocate the warehouse operative to process the consignments.
5.3.
Discussion
As Slack, Chambers and Johnston (2007) are stating, the uniformity of operations in the supply chain is essential for the supply chain performance. Simchi-Levi, Kaminsky and Simchi-Levi (2008) add thereto the major challenge in supply chain management regarding to replacing process with global optimization. Each stage of the supply chain optimizes its profit with no regard to the impact of its decisions on other supply chain changes. The objective is thus to coordinate the supply chain activities so, to maximize the overall supply chain performance.
Page | 37 of the limited time at the depots, 5 days maximum, it is questionable if a representative study has been conducted. It is desirable to gather and use data during a longer term in order to get a more reliable data set. Next to that, while this study is a combination of qualitative and quantitative data, it was leaning more towards the qualitative side. Shifting the focus more to quantitative data would result in an even more reliable study regarding time and costs savings. Apart from that, it would be wise to conduct a more thorough research on the flow of consignments at the depot itself taking into account the different performers of process.
5.4.
Further research
This study provides a decision model which is applicable for TNT’s Operations Europe. However TNT is a global operating organization and it is conceivable TNT will use the decision model worldwide. Though, it is questionable if the decision model is appropriate for global usage. This study used only data from European depots. Although all involved depots are European, they differ considerably. Therefore if TNT would like to use the decision model for non-european depots, further research should be necessary to acknowledge the applicability.
Furthermore, during the site visits many difference exists in the way of working on the shop floor. Every depot uses, according to them, the best way of processing the consignments. It would be advisable to conduct further research in comparing the depots regarding the work methods used on the shop floor. If a good comparison is made, with keeping in mind depot specific characteristics, the best aspects of all methods could be used to form one uniform method to realise a smoother flow of materials. Lately, TNT introduced projects using the Lean tool, which has a smoother flow of materials as a goal, and it is advisable to keep conducting projects based on Lean to increase efficiency of all processes (King, 2009).
Page | 38
6.
Bibliography
1. Abbelán, J., Masegosa, A.R., (2010) An ensemble method using creedal decision trees. European Journal of Operational Research, Vol. 205, pp 218-226.
2. Bianchessi, N., Righini, G., (2007) Heuristic Algorithms for the Vehicle Routing Problem with Simultaneous Pick-up and Delivery. Computers & Operations Research, Vol. 34, pp. 578-594. 3. Chow, G., Heaver, T.D., Henriksson, L.E., (1994) Logistics Performance: Definition and
Measurement, International Journal of Physical Distribution & Logistics Management, Vol. 24, No. 1, pp. 17-28.
4. Fitzsimmons, J.A., Fitzsimmons, M.J., (2008) Service Management: Operations, Strategy, Information Technology, International 6th edition, McGraw-Hill, USA.
5. Gleason, J.M., Barnum, D.T., (1982) Toward Valid Measures of Public Sector Productivity: Performance Measures in Urban Transit, Management Science, Vol. 28, No. 4, pp. 379-386. 6. Jaehne, D.M., Li, M., Riedel, R., Mueller, E., (2009) Configuring and operating global
production networks, International Journal of Production Research, Vol. 47, No. 8, pp. 2013-2020.
7. King, P.L., (2009) Lean for the Process Industries; dealing with complexity. Taylor & Francis Group, New York, USA.
8. Kotter, J., Rathgeber, H., (2006) Our Iceberg is Melting; Changing and Succeeding Under Any Conditions, Buttler and Tanner, Somerset, UK.
9. Lederer, P.J., Nambimadom, R.S., (1998) Airline Network Design, Operations Research, Vol. 46, No. 6, pp. 785-804.
10. McAfee, R.B., Glassman, M., Honeycutt, E.D., (2002) The Effects of Culture and Human Resource Management Policies on Supply Chain Management Strategy, Journal of Business Logistics, Vol. 23, No. 1, pp 1-18.
11. Mentzer, J.T., Konrad, B.P., (1991) An Efficiency/Effectiveness Approach to Logistics Performance Analysis, Journal of Business Logistics, Vol. 12, No 1, pp. 33-61.
12. Online TNT Annual Report 2009,
http://group.tnt.com/annualreports/annualreport09/index.html, last accessed 10 June 2010. 13. Schönsleben, P., (2007) Integral Logistics Management; Operations and Supply Chain Management in Comprehensive Value-Added Networks, 3th edition, Auerback Publications, USA.
Page | 39 15. Sink, D.S., Tuttle. T.C., de Vries, S.J., (1984) Productivity Measurement and Evaluation: What
is Available?, National Productivity Review, Vol. 4, No. 3, 1984, pp.265-387.
16. Slack, N., Chambers, S., Johnston, R., (2007) Operations Management, 5th edition, Pearson Education Limited, UK.
17. Slack, N., Lewis, M., (2008) Operations Strategy, 2nd edition, Pearson Education Limited, UK. 18. Wassan, N.A., (2006) A Reactive Tabu Search for the Vehicle Routing Problem. Journal of the
Operational Research Society, Vol. 57, pp. 111-116.
19. Woxenius, J., (2007) Generic Framework for Transport Network Designs: Applications and Treatment in Intermodal Freight Transport Literature, Transport Reviews, Vol. 27, No. 6, pp. 733-749.
Page | 40
Appendix A: Linehaul arrival times
Linehaul nr.
Arrival Time
Liverpool Belfast Kortrijk Paris
Airport Depot Bonneuil Créteil
1 0:45 7:30* 1:30 4:30 2:00 3:00 2 1:30 4:00 5:00 3:45 4:20 3 2:00 7:30* 6:22 5:45 5:00 4 3:00 6:30 6:15 5:10 5 3:30 6:30 6:30 5:45 6 5:00 7:30 6:30 6:30 7 5:30 7:30 7:00 7:00 8 6:00 8:30 7:00 9 8:45 7:15 10 9:00 7:30 11 9:00 % last 1/3 38% 100% 75% 45% 71% 50%
Page | 41
Appendix B: Cost components
Activity Cost component
Independent of activity Fixed costs
AM process Duration * costs per hour
Drive depot PUD area Duration * costs per hour
Distance * costs per km
Deliveries and pickups (road time) Duration * costs per hour
Distance * costs per km
Lunch €0,-
Drive PUD area depot Duration * costs per hour
Distance * costs per km
