Trunking Logistics for an E-grocer during Growth-A cost-centered approach in relation to service and centralization

Trunking Logistics for an

E-grocer during Growth

A cost-centered approach in relation

to service and centralization

by

M. D. Spekreijse

in Mechanical Engineering at the Delft University of Technology,

Committee chairman : Dr. ir. R. R. Negenborn

Supervisor DUT : Ir. M. B. Duinkerken

Supervisor Picnic : Drs. W. J. S. M. Bleeker

Reader 1 : Dr. ir. J. F. J. Pruyn

Reader 2 : Dr. B. Wiegmans

Delft University of Technology Department of Marine and Transport Technology Mekelweg 2 2628 CD Delft the Netherlands Phone +31 (0)15-2782889 Fax +31 (0)15-2781397 www.mtt.tudelft.nl

Student: M.D. Spekreijse Assignment type: doctoral assignment

Supervisor (TUD): M.B. Duinkerken Creditpoints (EC): 35

Spuervisor (EXT): W.J. Bleeker Report number: 2017.TEL.8108

Specialization: TEL Confidential: No

Subject: Trunking Logistics for an E-grocer during Growth

Picnic Supermarkets is the first online-only supermarket in the Netherlands and aims to compete with the existing supermarket giants by distinguishing themselves in service. This service, however, requires top performance every link in their supply chain. Picnic is interested in insights how their outbound logistics network, from their fulfilment centre towards their cross-dock centres, performs and what can be done to meet their performance targets for the coming years. Their main logistic activities are spread out over two branches: order fulfilment and distribution. In the fulfilment centre in Nijkerk articles are procured and approximately 100-150 order pickers support all governing processes. Distribution manages the home delivery or these orders and currently these activities take place in the cities of Amersfoort and Utrecht.

Picnic's trunking network, the outbound logistic activities from their fulfilment centre towards the cross-dock at distribution, needs a review on its current performance and if there are any

opportunities to its design to reduce costs for the following years. Also Picnic is still exploring how their future supply chain should look like, they are also interested in the impact of supply chain centralization on the performance of trunking.

The aim of this assignment is to identify potential improvements in the current trunking network. A model must be designed and implemented for the assessment of the designed alternatives on their performance under current demand as well as in future scenario's based on growth of the market. The report should conclude on how trunking should be managed in perspective to performance and the degree of supply chain centralization from a strategic point-of-view.

Studying relevant literature as well as the current processes at Picnic, developing and implementing a model, verification and validation of the model, experimenting, presenting solid conclusions and recommendations and reporting the research work are all part of this assignment.

The report should comply with the guidelines of the section. Details can be found on the website.

Dr. Rudy Negenborn

Section Transport Engineering & Logistics Dept. Maritime & Transport Technology Delft University of Technology

The switch of retail towards the online channel started in the 1990s and electronic commerce or E-commerce was born. Businesses like Amazon started venturing online by selling books, apparel or home electronics. Supermarkets also started their online channel, but slow internet connections made the experience inconvenient and time-consuming. The bulk of these e-grocers failed to at-tract customers and to cover their logistics costs in fulfilment and delivery. The bulk of the e-grocers were Traditional players using their current supply chain to fulfil their online operation. The online distribution activities resulted in relative high costs, because the complex delivery requirements and relative low consumer density diminish the already small margins.

Picnic is the first online-only supermarket in the Netherlands and believes that by distinguish-ing itself in service will attract customers to switch towards the online channel. From a logistic viewpoint this service is translated in free home delivery. During this research the Picnic supply chain will be considered the benchmark or Baseline for an online only e-grocer. See figure 1.5, this research will focus on the process of Trunking; the distribution of orders between the Fulfilment Centre (FC) and the cross-dock facilities called Hubs.

Delivery Outbound logistics

Picnic Fulfilment Supply Chain

Inbound logistics Wholesale /

Producers

Fulfilment Centre

Procurement _Trunking Hub Distribution Customer

Last Mile Outbound

E-Grocery Supply Chain

Inbound

Wholesale Fulfilment

Centre

Procurement Trunking Hub Distribution

(Cross-dock) Customer

Fulfilment Trunking Cross-dock _DeliveryHome

Figure 1: The Picnic Supply Chain consists of the steps Procurement, Fulfilment, Trunking and Distribution

The unique business model of being an online-only grocer with free home delivery leads to the lack of understanding how the Trunking network should be designed during growth. Within this research the design of Trunking during growth is strategically approached in relation to the degree of centralization for the FC and covering the supplementary expenditures of free home delivery (from now on referred to as Service). The analysis will be limited to Trunking, starting at the FC outbound flow and ending at Hub cross-docking. This modular approach of the supply chain will be respected, furthermore transport is carried out by a Third-Party Logistics Provider (3PL) and the impact of changes to customer buying behaviour is left out of the scope. The objective is to find indicative results and not exact solutions.

In order to achieve this goal, literature and the processes found at Picnic were analysed and four alternative FC-Hub designs were created. The alternatives consider a basic set of improvements, a temporary versus fixed Hub, the choice of carrier for Trunking and the degree of automation. An excel-based spreadsheet model was created, which modelled Trunking dynamically over the course of 5 years and compared the performance of the alternatives versus a benchmark. The model is based on a single FC-Hub network during the opening of a new market.

BASE

FLEX

FIX

HUD

A-HUD

Baseline

 JiT schedule

Flexible Hub

 Rent chilled cell  Buy mobile dock  Rent standtrailers

Fixed Hub

 Build chilled cell  Build dock Hub-dispatching  Implement HD operations Automatic Dock Hub-dispatching  Build automatic dock

Figure 2: Relations between alternatives

The experiments led to the conclusion that higher forms of centralization lead to higher savings when alternatives are applied. An e-grocer looking for opportunities in its Trunking network during growth considering costs, centralization and the service model the main driver is his financial po-sition. There are state-of-the-art solutions available for organizing a top performing network, but being a new player to a conservative market requires deep pockets if you want to stand out. Until the e-grocer has reached that position, it should play smart and look for opportunities which do not weight heavy on his available capital and should start with a decentralized approach.

This thesis is the result of research done on the outbound logistic processes of an online supermar-ket’s fulfilment centre. The focus lies on how this process is organized and how decision making on improving this network is influenced by factors like centralization, service and costs. Alone I would have never produced this thesis, therefore I would like to make my acknowledgements. First of all, I would like to thank my girlfriend, family and friends for their continuous support and patience for listening to my lengthy stories on online supermarkets, fulfilment centres and trucks.

This thesis has been carried out under the supervision of Mark Duinkerken. I would like to thank him for his continuous flexibility, straightforward feedback and commitment to always remain crit-ical of my work. Also I would like to thank Rudy Negenborn for taking up the responsibilities left by Prof. Lodewijks, I can only guess the tremendous extra work you have received.

For the past 9 months I have had the pleasure of combining this thesis with an internship at Picnic. Of all the options available, I have chosen to enter this immense dynamic environment were my findings were applied directly into practice. I cannot emphasize how fast-paced the past nine months have been.

I would like to thank my mentor Wybe-jan Bleeker for giving me the freedom to carry out my research. You emphasized the importance of hands-on experience in order to fully understand the problem. The opportunities you have given me made my Pinic experience lasting, I truly felt part of the team. Also, I would like to thank Frederik Nieuwenhuys for making me realise the importance of focus in work and life. Now I know I can never be too ambitious.

M. D. Spekreijse Delft, March 2017

Abstract iii

Abbreviations ix

Definitions xi

1 Introduction 1

1.1 Research context: Online Grocers . . . 1

1.2 Research context: Picnic supermarkets. . . 4

1.3 Problem description . . . 5

1.4 Research Scope and Goal . . . 5

1.5 Research Questions . . . 6

1.6 Methodology . . . 6

2 Literature review 7 2.1 E-Fulfilment . . . 7

2.2 Outbound logistics . . . 8

2.3 Distribution network design. . . 11

2.4 Challenges and solutions for e-grocers . . . 13

2.5 Analysis of logistic systems . . . 15

2.6 Conclusion and discussion . . . 16

3 Trunking System and Process Analysis 17 3.1 Black Box Approach . . . 17

3.2 The PROPER Model . . . 18

3.3 Operational Analysis . . . 20

3.3.1 Operations in and around Trunking. . . 20

3.3.2 Key Performance Indicators . . . 23

3.3.3 Parameters Blocking Trunking Performance . . . 24

3.3.4 Current Trunking Performance . . . 25

3.4 Conclusion and discussion . . . 26

4 Synthesis 27 4.1 Evaluation and Interpretation of Analyses . . . 27

4.2 Suggested Improvements for Trunking . . . 28

4.2.1 Chilled Storage at the Hub . . . 28

4.2.2 Docking facilities . . . 29

4.2.3 Stand trailers . . . 31

4.2.4 Hub-dispatching . . . 31

4.3 Alternative modes for Trunking . . . 33

4.4 Considerations prior to Experiments . . . 34

4.4.1 Approach of experiments . . . 34

4.4.2 Stakeholder analysis . . . 34

4.4.3 Interpretation of Assessments . . . 35

4.5 Conclusion and discussion . . . 36

5 Model 37

5.1 Model Setup . . . 37

5.2 Input . . . 39

5.3 Calculations . . . 40

5.4 Key Performance Indicators . . . 43

5.5 Conclusion and discussion . . . 44

6 Evaluation versus the Baseline 45 6.1 Experimental plan . . . 45

6.2 Results. . . 46

6.2.1 Average Total Costs (ATC) . . . 46

6.2.2 Investment Return based on Cash-flow (IRC) . . . 48

6.2.3 Multi-criteria analysis . . . 48

6.3 Conclusion . . . 50

7 Evaluation versus the ATC KPI Target 51 7.1 Experimental plan . . . 51

7.2 Results. . . 53

7.2.1 Average Total Costs (ATC) . . . 53

7.2.2 Investment Return based on Cash-flow (IRC) . . . 53

7.2.3 Multi-criteria analysis . . . 55

7.2.4 Post-processing . . . 56

7.3 Conclusion . . . 57

8 Conclusions and recommendations 59 8.1 Conclusion on research . . . 59

8.2 Recommendations for Picnic . . . 62

8.3 Recommendations for future research . . . 63

A Scientific Paper 65

B Background information on Picnic 73

C Analysis on Model Input parameters 79

D Operational plan alternatives 87

E Background information on Model 93

F Multi-criteria Analysis Setup 101

G Model Output 105

H Interviews 113

Table 1: Abbreviations used during research

Abbreviation Meaning

3PL Third-party logistics provider

ABC Activity-based costing

AHP Saaty Analytical Hierarchical Process

Amb Ambient

ATC Average Trunking Costs

BCA Benefit-cost analysis

CAPEX Capital expenditures

ChF Chilled/Frozen

DD Distribution density

DMS Distribution Management System

DPF Dispatch frame

DSA Delft Systems Approach

ePV Electric Picnic Vehicle

FC Fulfilment Centre

FIFO First in, first out

FTL Full truck load

HuD Hub-dispatching

IRC Investment return based on cashflow

JIT Just in Time

KPI Key Performance Indicator

LIFO Last in first out

MCA Multi-criteria Analysis

MPP Master Planning Process

P2A Peak-to-Average

PROPER Model Process-Performance Model

TAT Turnaround Time

TCA Total cost analysis

ULD Unit Loading Device

WMS Warehouse Mangement System

As not all readers are familiar with the terms used within this research the following table holds the fundamental terms used for this research. It is also advised to consult Appendix B.1 for the nomenclature on the Picnic Supply Chain.

Table 2: Terms used during this research.

Term Definition

KPI Target Cost target set by Picnic per link in the supply chain in order to save

costs over the whole supply chain to cover for services like e.g. free home delivery.

Benchmark Reference case used in experiments

Case study The situation found at the research’s sponsor Picnic

Alternative Newly designed FC-Hub configuration which will be assessed

within the experiments.

Trunking The process of distribution between warehouses

Assessment Interpretation and evaluation of model outcomes by applying a

multi-criteria analysis

Baseline Current situation found at Picnic

1

Introduction

This chapter introduces the research context, motivation and discusses its goal, scope and method-ology. The first section contains a general introduction to online supermarkets, e-commerce and the Dutch supermarket business. Thereafter, this research’s sponsor Picnic is introduced and their interests in relation to this research’s subject are explained. Section 3 discusses how these interests motivate this research and the 4th section contains the set scope and goals. The chapter is con-cluded by the research questions and -methodology which are this thesis’ roadmap.

1.1. Research context: Online Grocers

Online supermarkets were firstly introduced around the beginning of the millennium, the bulk of these failed to survive as slow internet connections made the customer experience inconvenient and time-consuming [31]. Apart from failing to create customer loyalty, e-grocers faced high logistic costs in storage, fulfilment and delivery. Online grocery market players can normally be categorized as either Traditional players, Online-Only players or Category expansionists[13]:

• Traditional players are existing retailers who develop a dedicated e-commerce business unit

next to their brick-and-mortar activities1, which means that their focus is on omni-channel

retailing (example: Albert Heijn (NL), Walmart (US) ). Some choose to outsource their e-commerce logistic operations, meaning they have initial no investment costs in infrastructure and capabilities (example: Whole Foods (US) ).

• Online-Only players are divided into four sub-categories:

1. Pure players dedicate all their resources to their online channel, which is their only chan-nel of operation (example: Ocado (UK)).

2. Concierge Services have a network of shoppers who are detached to deliver groceries which are picked in physical stores, resulting in a flexible network with no inventories (example: Shopwings (DE)).

3. Producers delivery straight from (relative small) suppliers, aiming for the more organic-and healthy-centric customers (example: FreshDirect (US)).

4. Niche / Category specialists focus only on one type of product and aim to offer unique assortments (example: HelloFresh (EU)).

1_{The term Brick-and-Mortar refers to the use of physical stores.}

• Category expansionists already have a global (online) presence and are in charge of a devel-oped IT, logistics and/or e-commerce network. These players could expand to online grocery shopping by using their existing business model (example: AmazonFresh (US)).

Brick-and-Mortar Supply Chain

Wholesale / Producers

Distribution

centre Supermarket Customer

Procurement Purchase

at store Distribution

Brick-and-Click Supply Chain

Wholesale / Producers

Distribution

centre Supermarket Customer

Procurement Home Delivery Supermarket Supermarket Distribution Customer Pickup Online Purchase Customer Purchase at store

Online-only Supply Chain

Wholesale / Producers Distribution centre Customer Procurement Home delivery Customer Customer

Pickup point Pick up Purchase

3rd Party Home delivery

Figure 1.1: Commonly used supermarket supply chain concepts per business model.

Most of the first e-grocers were traditional players who started the brick-and-click concept: orig-inal supermarket supply chains were extended with an online home delivery service. The online distribution activities resulted in relative high costs as the complex delivery requirements and rela-tive low consumer density diminished the already small margins. Especially the "last mile" delivery

concept2turned out to be a dominant cost factor[25]. The different delivery models employed by

e-grocery players can be divided as follows:

• In store Pick-up Points allow Brick-and-Click retailers to reduce the time spend in the super-market to the minimum, with relative small initial investments.

• Stand alone Pick-up Points are a more flexible answer to in store points as these points can have longer business hours and offer a more convenient solution for consumers living closer to these points.

• Home Delivery is the preferred option by consumers, but retailers do not favour this option as of the high initial investments; consumers are also not willingly to pay high delivery fees[38].

• With Outsourced Home Delivery a third party takes the grocery delivery in his existing net-work, which diminishes a lot of investments costs but the grocer loses control over its delivery process.

These different supply chain concepts are visualised in figure 1.1. To make use of the logistic net-work already in place, Brick-and-click grocers fulfil their customer orders in-store; however, some larger players also have fulfilment- and distribution centre’s which combine their off- and online logistic activities.

Supermarkets and E-commerce

Electronic commerce or E-commerce has started to rise in the 1990s, businesses like Amazon started online businesses selling apparel, books or home electronics. Currently the E-commerce market has evolved to a 24-hours economy, allowing consumers to buy almost anything from the web. Online grocery sales , however, have been lagging behind when compared to other product categories, see figure 1.2. This shows that there is a challenge in covering the gap between these product categories. The Netherlands is an ideal market for e-commerce businesses: 96% of the population is con-nected to the internet and 71% of the internet users shop online. Also the high population density in the Netherlands allow for the set-up of an cost-efficient distribution network[15]. The Dutch E-grocery market, however, is relatively immature with an online share of 1.5% of the total E-grocery retail revenue when compared to the UK (4.4%) and France (3.6%).

0% 15% 30% 45% 60% Groceries Cosmetics Books Home electronics Clothing and footwear

E-commerce sales of total sales France Germany United Kingdom Netherlands

Figure 1.2: E-commerce as a share of total market sales per product caterogy within the France, Germany, UK and the Netherlands[13][3].

Dutch Supermarkets

Grocery shopping is an accessible commodity in the Netherlands, according to the analysis of Rabobank [27] Dutch households have on average four different grocers within a close proximity and they do about two to three visits a week. The average distance towards a large supermarket in the Nether-lands is 900 meters[6]. Figure 1.3 holds figures on the Dutch grocery market of 2015. The left graph of figure shows the market share per player in total market revenue. The total market value is ap-proximately 33.8 billion euro’s and is mainly divided by large players as Albert Heijn (34%), Jumbo Group (20%), German Discounters like Aldi/Lidl (17%) and further are contested by more local gro-cers like Plus, Coop, Hoogvliet and Deen[13]. The right figure shows the current division of the online segment. Albert Heijn is dominant with a market share of 53% and the rest of the market is divided by numerous small players carrying out local delivery concepts.

In the first years of Dutch E-grocery development, the market leader Albert Heijn chose to charge relative high fees for home delivery in order to cover distribution costs; due to the lack of serious competition the market experienced a slow initial growth. Between 2013 and 2014; however, the market grew with 55% from 290 to 450 million euro’s; this was the result of multiple traditional su-permarkets, like the Jumbo Group, to start venturing online seriously. Most retailers favour the

Figure 1.3: Figures on the 2015 Dutch Grocery market size[13]

pick-up point option for last-mile fulfilment, this is parallel with a 680% growth of pick-up points in the Netherlands; it is still under debate whether this will be the most ideal solution for the last-mile fulfilment on the long run.

1.2. Research context: Picnic supermarkets

Picnic Supermarkets is the first online-only supermarket in the Netherlands and aims to compete with the existing supermarket giants by distinguishing themselves in service. Picnic has launched their online-only supermarket in September 2015. Customers can order groceries in the app-store up to 22:00 for the next day. Their main logistic activities are spread out over two branches: fulfil-ment and distribution. Currently the supermarket has opened its doors to the regions of Amersfoort and Utrecht, but it intends to upscale rapidly in the coming months. Being the sponsor of this re-search, the operations observed at Picnic will be used as the case study for a starting- and growing online-only e-grocer.

Figure 1.4: Picnic Supermarkets logo (Picnic Supermarkets, 2016)

Picnic Supply Chain Model

As a pure-player, Picnic handles it supply chain operations similar to the online-only player supply chain concept found in figure 1.1. The supply chain consists of the following steps: Procurement, Fulfilment, Trunking and Distribution, see figure 1.5. Picnic has decided its supply chain should be a modular supply chain. Each link can be considered isolated and this grants more flexibility in its design. This flexibility is, however, limited to the organization of the link itself, the in- and outputs are pre-defined to sustain a stable supply chain.

In their Fulfilment Centre (FC) in Nijkerk articles are procured and approximately 100-150 order pickers support all governing processes. The logistic steps from the FC point of view can be divided in Inbound- and Outbound logistics. Inbound logistics consists of all incoming streams of procured articles and the outbound represents fulfilled customer orders which are dispatched by truck from the FC and cross-docked at the hubs. The last mile tarts at the Hub and this process is called Distri-bution. Distribution manages the home delivery or these orders and currently these activities take place in the cities of Amersfoort and Utrecht. From now on the Outbound logistics process will be referred to as "Trunking".

Delivery Outbound logistics

Picnic Fulfilment Supply Chain

Inbound logistics Wholesale /

Producers

Fulfilment Centre

Procurement _Trunking Hub Distribution Customer

Last Mile Outbound

E-Grocery Supply Chain

Inbound

Wholesale Procurement Fulfilment _Centre Trunking _(Cross-dock)Hub Distribution Customer

Fulfilment Trunking Cross-dock Home Delivery

Figure 1.5: The Picnic Supply Chain consists of the steps Procurement, Fulfilment, Trunking and Distribution

1.3. Problem description

Having an unique business model, being an online-only e-grocer with free home delivery, Picnic is interested in how their supply chain should be designed. This thesis will consider Trunking, the processes of [20] and Distribution [37] have already been researched and Procurement is currently under research. Picnic lacks the understanding how their current Trunking should be organized during growth. They have interest in how strategic decisions on the distribution network design and their service model influence their decision making.

The motivation for this research is therefore three-fold. First, Picnic believes that their current design of Trunking is not fit for their position as a starting e-grocer and is interested in what op-portunities can be taken to improve its performance. Second, they are interested in the impact of the design of their supply chain network on Trunking costs and how this relates to their end-state supply chain. And finally, they are interested in the impact of choices within their service model on Trunking performance and if the costs following these choices can be covered. The scientific relevance of this research can be contributed to creating an understanding how an e-grocer should manage its Outbound logistics during growth.

1.4. Research Scope and Goal

This research will limit itself to the analysis of Trunking, starting at the FC Outbound flow up till cross-docking at the Hub. The modular supply chain approach from Picnic will therefore be re-spected and its setup will be left in tact. Also, the possible range of solutions will be limited to the current strategic choices of Picnic. At first, transport is carried out by a Third-party Logistics provider, Picnic has currently no interest and capital to take in these processes. And second, any possible changes in customer buying behaviour, the design of unit loading devices like the deliv-ery crates and -vehicles will not be considered. Last, a long term or strategic view on the design of Trunking is desired; Picnic is looking for indicative results on costs and not for exact solutions.

The goal of this research is threefold. It is aimed to analyse the current Trunking system and identify improvements which enhance performance, these insights should be gained from litera-ture and empirical research. Secondly, a model is to be created which can assess the effectiveness of these improvements to the Trunking system performance. Finally, this assessment should be

anal-ysed and concluded on how a Trunking network should be designed during growth by taking costs, supply chain centralization and service into account. As this research is to contribute to the long term decision making of the Trunking network, a strategic approach should be applied.

1.5. Research Questions

To reach the set goal, several research questions have been created. By answering each of these questions sequentially, the main question will be answered. This main research question is:

How should an e-grocer design its trunking network during growth in relation to costs, centralization and service?

This question is supported by the following sub-questions:

1. What are Outbound logistics for an e-grocer, which challenges do they face according to liter-ature and how can such a system be analysed on costs and performance?

2. How does the current Trunking network perform and how should it perform in relation to service quality?

3. How can the Trunking network be improved on its performance, by taking the learnings from literature and the analysis into account?

4. How can the Trunking system be modelled in order to assess the improvements on their per-formance?

5. How do the alternatives perform during growth in comparison to the desired performance?

1.6. Methodology

The following methodology will be applied, see figure 1.6. First a literature study (Ch.2) will be carried out to create an understanding of the outbound logistics. Second, the current state of op-eration at Picnic will be studied via empirical research (Ch.3). Thereafter the learnings from these two sections will be synthesised and recommendations on process improvements are made (Ch.4). Also the synthesis will discuss how the experiments should be set up and how decision making is approached.

Following the information gathering part of the research a basic model will be created and the alternative solutions will be implemented in this model (Ch.5). Thereafter the experiments will be carried out and are followed by a multi-criteria analysis (Ch.6 / Ch.7). The research will be con-cluded by answering the main research question, followed by recommendation for Picnic and fur-ther research(Ch.8). Literature study (Ch2) Basic model creation (Ch5) Apply improvements (Ch5) Experiments (Ch6/7) Validation/ Verification (Ch5) Conclusion & Recommendations (Ch8) Empirical research (Ch3) Synthesis (Ch4)

2

Literature review

This sections contains a review on literature on the subjects of e-grocery outbound logistics. The structure of this review is a top-down approach. The first section discusses the e-fulfilment models and how this relates to the research’s scope. Thereafter the multiple sub-processes within Outbound Logistics are discussed. The third sections contains a collection of approaches and learnings of Distribution Network designs. Thereafter a discussion on the challenges for e-grocers is held, why their online market share is lagging behind. Finally the different methods to analyse logistic systems on costs are listed and a choice of framework is made.

2.1. E-Fulfilment

In an e-fulfilment supply chain, orders are fulfilled from online sales. In comparison to offline re-tailers, "e-tailers" are able to save costs for running offline stores; however, their success is heavily dependent on customer acceptance. People can buy everywhere on the internet, customers are eas-ily lost to competitors if the service mode is not satisfactory enough; in logistics this is approached by enhancing the distribution- and delivery model. These models lead to extra expenditures within e-fulfilment and are unique to the e-tailer’s business model[18][21].

E-fulfilment is carried out in a Fulfilment Centre (FC) and the difference to a Distribution Centre is that a FC sends their orders directly towards the customer. Three types of e-fulfilment structures can be identified [12]: dedicated-, integrated- and store fulfilment. In contrast to dedicated fulfil-ment, integrated fulfilment integrates the e-commerce business unit within the existing operation and store fulfilment fulfils the orders within the existing retail stores. Brick-and-Click supermarkets tend to the latter two models, it is argued that in terms of performance they are inferior to dedicated fulfilment[18]. Dedicated E-fulfilment can be split up in four supply chain stages [2]:

• Sales are all processes and services which are in direct contact with customers. In E-fulfilment these orders are placed on websites or mobile applications.

• Purchasing are the upstream logistics of the warehouse. Articles are procured from suppliers and depending on the product variety the FC inbound has one or multiple suppliers.

• During Fulfilment, inbound articles are stored in the inventory and orders are fulfilled by order pickers or automatic warehouse management systems. Fulfilled orders are then send towards customers directly.

• Outbound logistics consists of all operations that physically move the product from the ware-house towards the customer. The definition differs per player but it can be split up in Distri-bution (or Trunking), Transshipment and Home Delivery.

This research’s scope is bound to the activities in and around Trunking (section 1.5). Figure 2.1 shows which subjects within e-fulfilment will be discussed in the following chapters.

E-fulfilment Purchasing Fulfilment Outbound logistics Sales Distribution Home delivery Out of scope In scope Transshipment

Figure 2.1: Graphical representation of subjects within and out of the research’s scope

2.2. Outbound logistics

Delivery logistics are one of the largest cost drivers of e-tailers. In order to increase operational effi-ciency the e-tailer is able to choose it’s "window of decision opportunity", see figure 2.2[34]. This ap-proach allows e-tailers to minimize their logistic costs in contrast to offline players. In e-commerce, quality of service is an important determinant for customer satisfaction[2]. If e-tailers want to run a sustainable operation the impact of service on their logistic costs is to be understood[26].

Figure 2.2: The online channel creates a window of opportunity for e-tailers [34]

Distribution

A distribution network is formed of one or multiple warehouses, transshipment hubs and a connect-ing transportation network [21]. Traditional retailers have a complex distribution network, consist-ing of multiple national- and regional distribution centres[35]; this results in a multi-stage distribu-tion network, see figure 2.3. Customers are divided on geographical locadistribu-tion, by their postal codes, and the given time windows for those locations[34].

Strategic decisions for an e-fulfilment supply chain cover the distribution network design: the choice of facility locations, -functions and transportation methods [28]. Facilities either have the function of fulfilment-, storage- or transshipment location. E-fulfilment networks have more flexi-bility in altering their setup up to the Last mile because the customer is completely decoupled. The main strategic decisions are made in the measure of Distribution Network Centralization versus lo-gistic costs, see figure 2.4. A higher form of centralization leads to lower costs in fulfilment, but higher costs in distribution[18].

4 Home delivery 3 Regional DC 2 National DC 1 Suppliers

Figure 2.3: An example of a four-stage distribution network

Transshipment

Transshipment facilities handle orders between two different carriers and may function as inter-mediate storage within the distribution network. Operational efficiencies can be achieved when freights can be consolidated with other shipments[16]. Depending on the exact function, these fa-cilities care called cross-docks, in-transit merge- or distribution centres:

• In-transit merge centres are stationed in networks which distribute products made up of multiple components. These components arrive from different suppliers and are packaged or assembled for the customer. The consolidation process is unable to start before all compo-nents have arrived.

• Distribution centres are intermediate warehouses which storage inventory between different stages to reach higher transport efficiencies.

• Cross-docks are intermediate facilities in distribution networks which act purely as a transfer location and are characterized by having no or little inventory. The concept allows integration of intermediate distribution transportation network nodes: inbound shipments are sorted on destination and directly transported to their destined outbound dock[19][33]. Smaller ship-ment between multiple suppliers and/or recipients can be consolidated such that full truck-loads are realised on at least one side. Cross-docking is typically carried out in the transporta-tion of perishable products, frozen foods and other refrigerated products as intermediate stor-age inside non-chilled docks is prohibited.

Figure 2.5: Schematic representation of a cross-docking hub[33]

Full truckloads contribute to the economics of transportation because of a reduced number of truck deliveries and traffic, however this requires some deliveries to make detours and to extend transport distances[4]. In contrast to point-to-point delivery, cross-docking increases staffing costs due to double handling times. The transshipment processes should be synchronized such that the intermediate inventory level can be kept at a minimum level. Different types of cross-docking ter-minals can be characterised with:

• The location of the cross-docking terminal can be either (1) close to the suppliers, (2) close to the customers or (3) the centre of a hub-and-spoke network. The first option is attractive in the case of multiple suppliers and a single customer, the cross-dock then functions as a

consolidation warehouse. The second options is attractive for a single supplier and

multi-ple customers, resulting in a break-bulk warehouse. The last option is of interest when the transshipment hub facilitates multiple directions in both in- and outbound, called a mixed

warehouse.

• For the intra-dock transportation shipments are usually transported by (1) transport vehicles e.g. forklift trucks, (2) an automated belt conveyor system or (3) an automated shuttle storage and retrieval system.

• Additional to the sorting- and consolidation processes, value-adding services may be carried out during cross-docking. These services range from adding supplementary labels for inter-national shipments or re-organizing the deliveries for ’just-in-sequence’ unloading.

2.3. Distribution network design

Three different levels of decisions can be distinguished when designing or managing distribution networks. These different levels differ on their time-scale and impact on performance: the strategic, tactical and operational level, see figure 2.6.

Operational

Tactical

Strategic

Choice of facility locations

Operational change

Capacity- and transport management

Planning Scheduling

Figure 2.6: Strategic-, tactical- and operational decisions in distribution network design

Strategic decision making

Long term and high capital investments decisions fall under strategic decision making, changes will take months to years to implement and will have a serious impact on the overall performance of the network. Strategic decision making largely concerns the exact locations of the facilities, the amount of stages within the distribution network and the function or design of each facility. The general design objective is to have the lowest total transportation distances, a dominant cost factor within the outbound logistics.

Tactical decision making

Tactical decisions for distribution networks mainly cover organizational changes, inventory- and distribution management. Organizational changes require a planning horizon of a few weeks to months, like a change to the production schedule. Inventory management decides whether the network will have storage facilities and which nodes will represent these. Distribution management decides on how outbound truck dispatching is managed in order to keep the combined inventory-and transportation costs to the minimum[7][8].

Operational decision making

The decision making around operational management is for leading the day-to-day operations. This will mainly consider choices on the planning and scheduling of workforce, trucks and deliveries.

Facility locations & functions Storage at warehouse Storage at distribution centre Storage at retailer Organization of operation Planning and scheduling of production and transport

Strategic Tactical Operational

Distribution management

Figure 2.7: Different decisions to make during network design

Distribution Network Design Considerations

The following approaches to the design of distribution networks can be found in literature. These case studies will be considered to be applied in this research.

• Max Shen and Qi [22] discus whether the integration of these different decision levels is viable. Long term decisions on facility locations influence inventory- and shipping costs. A simulta-neous approach of strategic- and tactical decision making is promoted, as the total costs will be taken into account. If a sequential approach is chosen it may result in sub-optimality of cost-performance.

• The work of Torabi et al. [34] has shown that customer orders are efficiently distributed from, maximum, the two most proximate fulfilment centre’s, as transportation is a dominant cost factor. Supplying from a single fulfilment centre is more cost-efficient for transportation, but strategic decision like split inventory or temperature zones may result in an efficient distribu-tion from two fulfilment centres.

• The work of Dowlatshahi [14] discusses on how to design logistic networks. A logistic system is build of sub-systems: isolated aspects which influence system performance and thus are critical design factors when developing effective operations. In their analysis they used the following design aspects: the transportation model, total costs, transport means, ULD usage and routing. In the design process the identified aspects should be taken as design KPIs in order to measure the design’s potential.

• A more top down approach is presented in Jayaraman and Ross [19]: a ’Production, Logistics, Outbound Transportation’ or PLOT design system is presented. The design system consists of two types of decision making phases: a strategic based stage were the network is estab-lished and an operational stage which runs the network efficiently. The first stage handles the location-allocation problem of placing warehousing facilities, distribution centres and cross-docks in order to minimize the total of warehousing, transportation and inventory costs. The

second stage is an executional phase and is meant to effectively deal with the system which has been created in the first phase. This stage is driven by minimizing the total costs of trans-portation, transshipment and distribution.

• In the work of Palm et al. [24] it is argued that the introduction of logistic constraints to pro-duction scheduling could lead to the optimization of the outbound transport. In their work they distinguish efficiency in both economical-, as total costs, and ecological terms, as ex-ternal effects caused by the transport. Solutions can be sought on operational-, tactical- and strategic level:

– The short term solutions focussed on rescheduling the production planning and trans-port. Higher efficiency can be attained when the planning is adjusted to external im-pulses like traffic and stochastic demand.

– Tactical decision-making resulted in considering other means of transport to increase the total carrier capacity and outbound throughput. Rail or ship transport is generally considered more efficient than truck driving as their carrying capacity is larger.

– A strategic implementation was proposed on integrating production- and routing plan-ning, resulting in a capacity-oriented production planplan-ning, please find the example in figure 2.8. Multiple batches for the same destination which are produced over the day, should be produced in a single batch, this only holds if the total demand does not exceed the carrier capacity.

1 2 3 1 2 3 4 5 A1 A1 B1 C1 C1 A2 A2 A2 B2 B2 B2 A A A A A B B B B C C Convential production planning Capacity oriented production planning

1

2

Figure 2.8: A capacity oriented production planning (2) results in efficient outbound transport

2.4. Challenges and solutions for e-grocers

The introductory chapter showed a significant different in online sales share for groceries in com-parison to other product categories. E-grocers are lagging behind and these differences can be ex-plained by the challenges which they face in practice:

Preservation of fresh and frozen articles during distribution

Grocery baskets contain fresh and frozen articles which decay when exposed to higher temperatures over time. By containing these orders within isolated boxes this period can be extended to a few hours, but this will still lead to compromises within the supply chain.

• Either the orders are delivered Just-in-Time (JiT), a methodology which requires the chain to be tuned so that every link receives what they need just in time. JiT allow transshipment locations to have no storage function, which lowers costs but makes the system sensitive to disturbances[17].

• Or the supply chain has intermediate chilled storage facilities in transport and/or transship-ment. This options is viable for players who favour freight consolidation as higher transport efficiencies can be made. However, this leads to higher transport costs and capital invest-ments in the transshipment locations.

Product vulnerability

When Fresh foods are unpacked to consumer units they become vulnerable to impacts during

pick-ing, packing and transport1. From a logistic perspective, grocers should emphasize on minimizing

the transport distances and the degree of product handling to diminish the risk. Also improvements on product packaging or creating protective layers in the picked orders may prevent any damage.

The Last Mile

The Last Mile can be realised as Home Delivery or Pick-up and again Home delivery is split in at-tended and unatat-tended delivery. The first model leads to comprises in either logistic costs or cus-tomer experience. The cuscus-tomers needs to be at home during delivery and a high accuracy in plan-ning is required to satisfy their needs.

Unattended delivery has showed to be successful for the Swiss e-grocer Leshop. They have im-plemented unattended home delivery model in which they deliver the groceries in cooled delivery boxes which are stationed at the customers homes. In the work of Hubner et al. [18] it is argued that this solution may be succesfull in high GDP countries like Switzerland, but that regions in lower GDP countries may be prone to theft or vandalism. Also this delivery model requires either the customer or the grocer to invest in storage facilities.

Pick-up points require the customer to gather their fulfilled orders at a centralized Hub. In com-parison to home delivery this gives the customer more flexibility when he wants to pick up his gro-ceries, but his can be considered a lesser service. Figure 2.9 shows the difference in order charac-teristics for Home Delivery and Pick-up[25]. It also shows that the last mile is more expensive for Home Delivery and that these expenditures can be covered by asking delivery fees.

Figure 2.9: Difference in cost-structures between home delivery and pick-up[25]

Free delivery and returns

The current trend in e-commerce is set at free delivery and even some are offering free product returns. The first options was discussed in the previous section, but for e-grocers return policies do not work, for two reasons:

• In contrast to e.g. selling books, food returns cannot be resold and are therefore direct losses. These rejections occur when product quality does not match the customer’s expectations, either the product is decayed or damaged during transport.

• The reverse logistics around food returns are complex and will result in high operational costs for the e-grocer. For example, in the Netherlands it is not accepted to send food via the com-mon mail system.

Relative low order value per cubic meter

Supermarkets are competitive and stunt frequently with prices to undercut the competition and therefore the grocery business can be called "a game of small margins". Grocery shopping baskets hold up to 60 times more articles than a non-grocery online order[18]. Also the average supermarket article has a relative larger product volume and the small margins result in a relative low value per cubic meter[5]. If the organization of free delivery logistics for e-commerce is already challenging, supermarkets face a big challenge. Solutions to enhance the value per order is considered out of scope (section 1.5) and will not be discussed here.

2.5. Analysis of logistic systems

Logistic systems can be analysed on their performance from multiple perspectives. The research’s scope (section 1.4) stated that the analysis should be carried out from a cost-perspective. Cost analyses are executed to understand the system mechanics by identifying which resources drive performance[1]. Supply chains are typically analysed from a holistic viewpoint, approaching the system as a whole and aiming of minimizing the system’s global costs. Models to analyse costs of a system can either exist to implement a systematic method to monitor and analyse costs for a system or to analyse the impact of specific decisions, practices or future interventions. In literature three different frameworks are usually applied for these analyses:

• Traditionally systems are assessed on their costs via Cost-benefit analyses, however, these are not properly adjusted for logistic systems as not all system impacts are automatically taken into accountChristopher [9].

• The works of Christopher [9], Corla-souza et al. [11] and Shapiro [32] promote the Total cost

analysis (TCA) approach in order to compare transport alternatives within a logistics system.

The purpose of TCA is to visualise the change costs by the changes to the system, the design choices can be linked to costs in incremental terms.

• Activity-based costing(ABC) provides an advantage when the costs of services are to be iden-tified [10]. ABC seeks the cost-drivers along the supply chain and assigns costs to each level of activity, giving a clearer picture of the true total costs.

For this research it has been decided that TCA will be applied as a analytical framework. TCA is deemed to be the right choice because it is better suited for transport systems and for this research it is unnecessary to allocate costs to certain activities (ABC). The analysis is carried out in multiple phases:

1. System and Process analysis

The logistic system should be analysed from a holistic viewpoint by considering its role as a sub-system, followed by a more in-depth research on the core operational processes. This should identify the system KPIs and its main cost drivers while respecting the relation to the whole system. These drivers are influenced by cost parameters, for further analysis these pa-rameters should be quantified in their relation to the current system performance.

2. Creation of alternative modes

Following the analysis of the current state of performance and the interpretation of the pro-cess, improvements to the logistic system are worked out. From these improvements different future alternative modes will be designed. Also the definition of the decision making is worked out to identify which values are of importance for the future system.

3. Modelling and Evaluation

To assess the impact of the alternative modes on the system a model is created. First this model should be compared to the current system state to calibrate its parameters. There-after the alternative modes are implemented. The link between the cost parameters and the system costs is mapped with a sensitivity analysis. Finally the alternative modes should be ex-perimented with in relation to the current state. By evaluation these assessments a conclusion and advise can be made.

2.6. Conclusion and discussion

Outbound logistics are the FC downstream activities and can be split up in Distribution, Trans-shipment and Home Delivery. Distribution is the transport between warehouses, transTrans-shipment locations and the customer, the design’s target should be to minimize total transport distances. In transshipment facilities cross-docking and/or consolidation of freights take place to increase trans-port efficiencies.

The network’s design may lead to one or multiple stages. It was found that the design of such a network can be approached from a Strategic-, Tactical- and Operational level. There are design frameworks present which promote to combine these different decision levels, but it depends on the pursued goal whether this is desired. If the considerations found are compared with the research’s goal, it is interesting to combine a strategic and tactical approach: the combination of the network stages and the design of each facility. The measure of network centralization is strongly related to its performance. Therefore the effect of centralization on transport should be investigated to contribute to the strategic question how the complete network should be designed.

The organization of an e-grocery supply chain result in extra challenges. These can be assigned to the transport of refrigerated articles, product vulnerability, the last mile and the value per order. For the outbound logistics it is important to make the right decisions around the challenges to lower overall costs while keeping the customer satisfied. In order to lower these costs a analytical frame-work should be applied. This research will apply the Total Cost Analysis (TCA) frameframe-work. The TCA methodology will be used to analyse the current system and to compare the cost-performance of alternative modes. By taking these learnings from literature into account, the following chapter will contain the system and process analysis of the system at Picnic.

3

Trunking System and Process Analysis

This chapters discusses the Trunking network found at Picnic and analyses its performance. It is aimed to understand the system’s workings, its current and required performance. The supply chain may have a different nomenclature than readers are used to, therefore it is advised to consult Ap-pendix B.1 to avoid any misinterpretation. The first two sections discuss Trunking as a sub-system of the supply chain by applying the Delft Systems approach. Thereafter the operational activities within and around Trunking are discussed. The final section will conclude on the current state of performance found at Picnic.

3.1. Black Box Approach

A system is greater than the sum of its parts, therefore the Trunking system will be analysed as part of the complete system. This section will start with the analysis of the supply chain model with the black-box approach. A black box is an abstraction of a system where only the inputs and outputs are considered, leaving the internal workings out of the scope; in the following sections the black box will be gradually opened to allow for a more detailed analysis.

Picnic Supply Chain

Groceries

App only store

Orders

Home delivery

Timeliness

Quality

Sustainability

Figure 3.1: Picnic supply chain as a black box

Figure 3.1 shows the complete system as a black box, where orders placed by customers are input and home delivery of grocery orders as output. The Picnic Supply Chain values Timeliness, Quality and Sustainability as their main process drivers:

• Timeliness demands on-time delivery at the customer, Picnic tries to distinguish itself by hav-ing a plannhav-ing accuracy within a 20-minute window. In order to achieve this, Timeliness is an important design value for every link in the supply chain. This demands from every link that it will not hinder its successors timeliness.

• Quality refers to the requirement of a stable process around the delivery of undamaged, pre-served articles. The distribution of fresh articles is a sensitive business, articles are easily dam-aged and lengthy operations allow frozen goods to defrost.

• Sustainability is required, because due to the immense competition players should have a sustainable approach to all operations. By promising free home delivery, timeliness and qual-ity it requires its supply chain to have the best cost-performance. Therefore Picnic has valued key performance indicators (KPIs) for every link in order to achieve these goals. These will be discussed in section 3.3.

3.2. The PROPER Model

Various tools can be used to decompose and analyse systems, and for this research the Delft Systems Approach (DSA) has been chosen. The DSA is promoted by Veeke et al. [39] for the analysis of in-dustrial systems and it focusses on abstracting the system into a material-, order and resource flow. Opening the black box the Process-Performance (PROPER) model becomes visible and it is shown in figure 3.2(a). The figure shows that the supply chain can be divided in three streams Orders, Articles and Resources, furthermore there are two controlling instances Management and Customers:

• The flow of Orders is the main input of the system. Customers place orders in the online store and is characterized by the set of articles, delivery location and delivery shift. The stream execute processes these orders and releases the orders towards the Supply Chain Operations. • The flow of Articles start at procurement, the inflow of articles into the FC, and ends when the orders are delivered at the customer. Within the supply chain operations two main processes can be found: Process & Control and Fulfil & Distribute. These processes will be discussed further in this section.

• Resources fulfil the deployment of required assets throughout the supply chain. These are mainly the Unit Loading Devices (ULDs) which carry the orders throughout the supply chain. These resources are discussed in section 3.3.

• The Management directly controls all supply chain operations, taking the results on Timeli-ness, Quality and Sustainability into account and steering the operation such that the mile-stone targets are reached. These values have been discussed in section 3.1.

• The requirements set by the management are created from Customer demand. Picnic be-lieves that its core business should be the fulfilment of a service, a part from fulfilling gro-ceries. These requirements set by Picnic will be discussed in section 3.4.

The following section will open up the Supply Chain Operations stream, discussing further how Process & Control and Fulfil & Distribute are managed.

Supply Chain Operations

By further zooming in the underlying processes are revealed in figure 3.2(b). Customers may place orders up till 22:00 the night before, this is the cut-off time of the online store. After all orders are processed they are released in the Master Planning Process (MPP). The MPP is the core-calculating programme which determines how the Electric Picnic Vehicles (ePVs) should drive to minimize dis-tribution costs. The MPP starts by calculating the exact amount of space each customer needs in an ePV. After calculation all orders are then put into production planning in a first-in-first-out (FIFO) order. The time between the MPP and final delivery is the window of decision opportunity for Pic-nic, see figure 3.3.

During this window of decision opportunity the Fulfilment, Trunking and Distribution pro-cesses are executed. These propro-cesses are controlled by the Warehouse- and Distribution

Manage-ment Systems (WMS, DMS). The WMS takes up control of the operations within FulfilManage-ment and

Trunking whilst the DMS controls the operation from the distribution side. These processes will be further discussed in the following section by opening the Supply Chain Operations (figure 3.2(b)) to an operational level.

Management

Execute

Use

Handled order

Resources

Delivered order

Process & Control

Fulfill & Distribute

- Articles - Delivery window - Hub Order fulfilment - Fulfill to groceries - Outbound by truck - Transshipment at Hub Employ resources - Totes - Dispatch frames - Trucks

Articles

Customers

(a) The PROPER model applied to the Picnic Supply Chain

Execute

Fulfilment Trunking Distribution

Warehouse Management System Distribution Management System Master Planning Use

(b) The governing processes within the Supply Chain Operations Figure 3.2: Analysis of the supply chain with the Delft Systems Approach

22:00 04:00 16:00 22:00

Cut-off time (t) MPP finished Cut-off time (t+1)

Delivery Window of decision opportunity

First delivery

Figure 3.3: Picnic’s window of decision opportunity

3.3. Operational Analysis

The previous section showed that in the PROPER model three major processes are found in the Supply Chain: Fulfilment, Trunking and Distribution. This section will discuss these processes on operational level. In order to respect the scope, the main focus will lie on Trunking and the neigh-bouring operational activities. Thereafter the KPIs and main cost drivers are discussed, followed by a decomposition which should identify the important cost parameters of Trunking.

3.3.1. Operations in and around Trunking

This section discusses the main operation of and any important processes which are related to Trunking. The process-flow diagram found in figure 3.4 should be consulted during this section and the photos in figure 3.5 can be used to comprehend the explanation.

Fulfilment

All processes within the FC fall under Fulfilment and are controlled by the WMS, in Appendix B.2 the layout of the FC can be found. The FC is divided in three different temperature zones: Am-bient, Chilled and Frozen. Articles are procured via suppliers and are received in the FC, split up per temperature zone and replenished in the inventory. Order Pickers pick articles to fulfil the cus-tomer orders, see figure 3.5(a). Picked articles are put into totes, these are red for ambient (Amb) orders and black for chilled/frozen (ChF) orders. After picking totes incomplete totes are put onto the consolidation stockpile and wait until the final articles have been replenished.

Complete totes are transported towards Dispatching. In dispatching totes are stored in Dispatch Frames (DPFs), see figure 3.5(b). DPFs are metal ULDs for the totes, in the current state there are 3-layered (18 totes) and 4L (24 totes) DPFs. The dispatching processes are isolated in Amb and ChF. After DPFs are stored with all orders they are ready for shipping, a set of dispatched DPFs destined for a Hub is called a shipment.

Trunking

Trunking is the transport of DPFs between the FC and Hubs. This transport is carried out by trucks from a Third-Party Logistics Provider (3PL), see figure 3.5(c). When dispatching has readied a ship-ment, the DPFs are loaded from the docks into the trailer. DPFs are equipped with wheels and there-fore they can be loaded manually. 40 DPFs fit into a single trailer formed in two rows of 10 DPFs deep stacked 1 high. The introduction of 4L DPFs allows Distribution to increase its efficiency, however, the current set of available trucks in the market are not fit for 40 4L DPFs. Therefore the system uses both 3L and 4L frames as they fit stacked upon each other within a trailer.

When fully loaded, trucks drive towards the destined hub(s). Hubs are designed in a minimalis-tic way and there are no docking capabilities present, therefore trucks are forced to unload the DPFs with the tailgate and the process flow is shown in figure 4.2(a). The unloaded shipment is put into storage until Distribution starts. Trunking also governs the return flow of DPFs after Distribution back towards the FC.

D is tr ib ut io n Tr u nk in g D M S W M S O rd er s - Ex ec u te R es ou rc es U se DP Fs To te s To te s/ DP Fs Fu lfi lm en t M PP R e ce iv e O rd e rs P la n Dis tr ib u tio n C o n tr o l In b o u n d Sc h ed u le O p er a tio n s C o n tr o l P ic kin g C o n tr o l Dis p a tc h in g C o n tr o l C o n tr o l Dis tr ib u tio n R e ce iv e / re p le n is h O rd e r p ic kin g Dis p a tc h C on so lid at e Lo ad t ru ck Dr iv e t o H u b U n lo ad tr u ck De liv e ry R e tu rn e P V Lo ad t ru ck U n lo ad tr u ck Dr iv e t o FC P la ce DP F o n d o ck Ch eck U n st a ck DP Fs St o ra ge Lo ad e P V U n lo ad e P V St ac k DP Fs St o ra ge F igur e 3.4 : O p e rat io n al act ivities wi thin F u lfilmen t, T runking and D istr ibution fr om a T ru n ki n g p erspect iv e

(a) Order picking at the FC (b) Storing totes in dispatchframes

(c) Trunking: Shipping towards the Hub

(d) Transshipment from truck to ePV (e) Home delivery with the ePV

Figure 3.5: The operational activities within the Picnic Supply Chain (Picnic Supermarkets, 2015)

Park truck Pick DPF Unload DPF

Pick empty DPF Load empty DPF Rise tailgate Lower tailgate No Prepare truck Finished?

(a) Truck handling by tailgate

Park truck Pick DPF Unload DPF

Pick empty DPF Load empty DPF Prepare truck Finished? No

(b) Truck handling via dock

Distribution

All processes which support The Last Mile from the Hub on, fall under Distribution. The Hub is a cross-docking facility from where the ePVs start delivery. Figure 3.5(d) shows that during cross-dock the DPFs are loaded into the ePV. Every hub has an unique delivery area and delivery is carried out in multiple delivery windows called Shifts, figure 3.5(e) shows the ePV in action. An ePV drives a single trip per Shift, carrying a set of Amb and ChF DPFs. During this research a single distribution round of an ePV will be designated as Trip.

3.3.2. Key Performance Indicators

In order to respect the Supply Chain Value Sustainability (section 3.1) and the modular supply chain setup (section 1.3), Picnic has decided that every link in the supply chain requires a specific per-formance target. This has lead to every process having a single KPI to target for. These KPIs are called the Average Fulfilment Costs (AFC), Average Trunking Costs(ATC) and Average Distribution Costs(ADC). These KPIs are the total costs made by the process divided by the unit of measure trans-ported within, so for Fulfilment these are articles and for Trunking/Distribution these are Orders.

KPI Target based on Service

The business model found at the case study grocer Picnic has chosen to guarantee free home de-livery. In order for to sustain this model, Picnic has decided to set KPI Targets for every link in the

supply chain. For Trunking it has been decided that the ATC has a KPI Target ofe1.00 per order. The

ATC is calculated by dividing the Trunking Costs (Ct) by the transported order volume (V ):

AT C =Ct V

The ATC KPI has been decomposed in figure 3.7, it is influenced by the development of the Order

Volume (V ) and the Trunking Costs (Ct). The Ctare the total costs made by the 3PL:

Ct= chr× T + ckm× D

These costs are split up in a time- (chr×T ) and a distance component (chr×D). The Total Time (T ) is

split up in time spend driving, handling trucks and idle time. Total Distance (D) is dependent on the

FC-Hub distance (d ) and the amount of Trunking Trips (nt t r). Figure 3.7 shows the decomposition

of the KPI and which cost parameters influence its performance, please find an analysis of these cost parameters in Appendix C.

Order volume development Distribution of volume Distribution density Peak loads Truck driving Hub / FC Operations Truck idle time

Distance FC-Hub Building expenses Material expenses ATC Trunking costs Order volume Total time Total distance Characteristics

Growth  Volume development

 Distribution density

 Volume distribution

 Peak loads

 Handling time

 Truck time division

 Truck driving  Hub – FC distance  Amount of trips Total costs Trunking costs Operational expenditures Investments Costs / order Capital expenditures Savings

KPI Cost drivers Cost parameters

3.3.3. Parameters Blocking Trunking Performance

The cost parameters defined in figure 3.7 are analysed in Appendix C on their behaviour and per-formance. Following that analysis and the observations made in the process (Section 3.1-3.3) the following can be stated:

• The combination of multiple delivery windows, the transport of ChF orders and the minimal-istic design of the Hub prevent Picnic to fill their trucks efficiently for small volumes. If the total volume for that day fits in a single truck Picnic still ships them separately, because ChF orders can only be exposed to warmer temperatures for a limited period.

• The analysis in Appendix C1 showed that the Order Volume is distributed unevenly over the days and shifts. This distribution is out of control and out of scope for this research. These disruptions influence Trunking in a negative way because it demands extra capacity for one or two days a week.

• While trucks can be efficiently filled with DPFs, Trunking’s performance is dependent on the efficiency of Distribution: the Distribution Density (DD), this parameter is discussed in Ap-pendix C.1. The analysis showed that during growth DPFs are filled inefficiently and observa-tions showed that it can be assumed that DPFs are on average for 85% full. Therefore it can be concluded that by using DPFs trucks transport empty spaces which could have been filled. • In Appendix C2 the distribution of truck time has been analysed and it showed that trucks

have waiting times at the FC. These waiting times have been quantified in figure H.1(a) and occur when the FC is delayed in production. The FC layout shown in Appendix B shows that there is a limited amount of space for dispatching and shipping. If there are no available trailers for loading, docks cannot made empty and FC production is stalled. Therefore both Trunking and Fulfilment are blocked in their performance by the unavailability of trailers at the docks.

• The analysis on truck driving in Appendix C2 resulted in the assumptions on average truck driving speeds found in figure H.1(b).

• Figure H.1(c) shows the result of analysing truck handling times at the FC and Hub. It shows that the (un)loading process by using the truck tailgate is time-consuming and blocking per-formance. 0 10 20 30 40 FC Hub M in ut es p er T ru nk in g Tr ip

Waiting Break Other

(a) Assumptions on truck idle times 0 20 40 60 80 20km 40km 60km 80km D riv in g Sp ee d

(b) Assumptions on Truck driving

0 15 30 45 60 FC Hub M in ut es p er F TL

(c) Assumptions on truck handling Figure 3.8: Results from the analysis on Trunking operations