Multi-Echelon Inventory Optimization for Fresh Produce

Authors: Saran Limvorasak and Zhiheng Xu

Advisor: Dr. Francisco Jauffred

Sponsor: A Mass Discount Retailer

MIT SCM ResearchFest May 22-23, 2013

Fresh Produce for Retail Business

• Product Freshness and Availability are key attributes for a company competing in grocery segment in retail business.

May 22-23, 2013 MIT SCM ResearchFest 2

Introduction

MIT SCM ResearchFest 3

Should a mass discount retailer add upstream

produce facilities into its current network?

• What are key benefits from an additional node?

• What are the impacts to supply chain networks?

• Analyze Top 21 fresh produce categories

• Develop a Predictive Model to compare a network with and without an additional facility

Thesis Questions:

Scope and Expected Outcomes:

May 22-23, 2013

Supply Chain Network Under Two Scenarios for Comparison

MIT SCM ResearchFest 4

Supplier Grocery

Distribution Center

Retail Store

Supplier Grocery

Distribution Center

Retail Store Fulfillment

Center

Hold Inventory Replenishment Frequency

Hold Inventory Replenishment Frequency

Average 7 times/week

Average 3 times/week

Average 3 times/week

Average 7 times/week

Scenario 1: Existing Supply Chain Network

Scenario 2: Supply Chain Network with Fulfillment Center

May 22-23, 2013

Risk Pooling

MIT SCM ResearchFest 5

Key benefit from moving inventory upstream is from Risk Pooling

• The concept of Risk Pooling is a powerful tool to address variability in the supply chain

• Benefit from having central warehouse is greater in a system in which demand has higher volatility

Supplier

Regional

Warehouse

Regional

Warehouse

Supplier

Regional

Warehouse

Regional

Warehouse

Central Warehouse

Network 1: No Central Warehouse Network 2: With Central Warehouse

Network 2 with Central warehouse has less Safety Stock and Average Inventory because

May 22-23, 2013

Methodology

MIT SCM ResearchFest 6

Supplier

Fulfillment

Center

Grocery

Distribution Center Retail Store

TrGDC

TrFFC TrGDC

TrStore

TGDC TStore

TFFC

Total supply chain cycle time “A total time which a product spends in the supply chain

from supplier until it is sold”

Inventory Dwell Time (T) : Average time which product is stored at facility

Transit Time (Tr) : Average transportation time between facilities

Safety Time (TSf ) : Safety Time in the supply chain captures the effect of demand volatility at Retail Store

STEP I: A Predictive Model

Scenario 1 Scenario 2

May 22-23, 2013

Methodology

MIT SCM ResearchFest 7

STEP I: A Predictive Model

May 22-23, 2013

Methodology

MIT SCM ResearchFest 8

STEP II: Simulations of the inventory levels in supply chain:

DOSFFC

DOSGDC DOSStore

• Focus on the inventory level at each inventory facility

• Relax assumptions on Inventory Policy by using current periodic inventory policy (R, s, S)

• Average Inventory Level and Days of Supply (DOS) are supply chain performance metrics

Supplier

Fulfillment

Center

Grocery

Distribution Center Retail Store

Scenario 1 Scenario 2

May 22-23, 2013

Methodology

MIT SCM ResearchFest 9

STEP II: Simulations of the inventory levels in supply chain:

May 22-23, 2013

MIT SCM ResearchFest 10

Average

Demand

(lbs)

Standard

Deviation

Enhanced

Coefficient

of Variation

Safety Time

(days)

Transit Time

(days)

Total Supply

Chain Cycle

Time (days)

1 Berries 15 30 1.97 (0.97) 0.3 (0.47)

2 Watermelons 57 62 1.10 (0.54) 0.3 (0.24)

3 Cherries 64 60 0.93 (0.38) 0.3 (0.08)

4 Mixed Melons 57 49 0.86 (0.34) 0.3 (0.04)

5 Stone Fruit 165 133 0.80 (0.32) 0.3 (0.02)

6 Strawberries 176 11 0.64 (0.31) 0.3 (0.01)

7 Citrus 195 88 0.45 (0.22) 0.3 0.08

8 Nuts-Snacks- 35 15 0.44 (0.21) 0.3 0.09

9 Grapes 317 117 0.37 (0.18) 0.3 0.12

10 Avocadoes 340 125 0.37 (0.18) 0.3 0.12

11 Potatoes 107 35 0.32 (0.16) 0.3 0.14

12 Cut Fruit 109 35 0.32 (0.16) 0.3 0.14

13 Apples 331 105 0.32 (0.13) 0.3 0.17

14 Mushroom 44 11 0.26 (0.13) 0.3 0.17

15 Mixed 151 47 0.31 (0.12) 0.3 0.18

16 Carrots 104 26 0.25 (0.10) 0.3 0.20

17 Onions 297 74 0.25 (0.10) 0.3 0.20

18 Lettuce 203 49 0.24 (0.10) 0.3 0.20

19 Tomato 434 86 0.20 (0.07) 0.3 0.23

20 Pkg Salads 292 57 0.19 (0.07) 0.3 0.23

21 Bananas 1,427 238 0.17 (0.06) 0.3 0.24

Product Category

Incremental / (Saving)Demand Characteristics

Total Supply Chain Cycle Time

• From 21 Product Categories, Total Supply Chain Cycle of 6 Product Categories in Supply Chain Network with Fulfillment Center is Less than existing network

• All Product Categories have reduction in safety stock

May 22-23, 2013

Enhanced Coefficient of Variation

MIT SCM ResearchFest 11

• Enhanced Coefficient of Variation (ECV) is created to measure the relative demand volatility

May 22-23, 2013

Enhanced Coefficient of Variation Break-Event Point

MIT SCM ResearchFest 12

Vendor replenishment frequency ECV Break-Event Point

1 time a week 0.45 2 times a week 0.63 3 times a week 0.76 4 times a week 0.88 5 times a week 0.99 6 times a week 1.08 7 times a week 1.18

• Sensitivity Analysis on Vendor replenishment frequency is tested to determine the break-event point for Enhanced Coefficient of Variation

May 22-23, 2013

Average Inventory in Supply Chain

MIT SCM ResearchFest 13

Scenario 1 Scenario 2 Incremental /

(saving)

1 Bananas 1,305,661 1,237,040 (68,622)

2 Avocadoes 301,327 280,422 (20,905)

3 Citrus 179,321 164,461 (14,860)

4 Apples 286,792 272,313 (14,479)

5 Stone Fruit 166,995 152,759 (14,236)

6 Grapes 277,742 264,727 (13,015)

7 Strawberries 167,591 154,646 (12,945)

8 Mixed 133,407 126,201 (7,206)

9 Cherries 71,453 65,699 (5,754)

10 Cut Fruit 97,094 92,149 (4,945)

11 Watermelons 65,781 60,959 (4,822)

12 Potatoes 94,608 90,232 (4,375)

13 Mixed Melons 61,672 57,376 (4,297)

14 Carrots 90,730 87,941 (2,789)

15 Berries 24,770 22,265 (2,504)

16 Pkg Salads 247,322 245,231 (2,092)

17 Nuts-Snacks- 34,900 32,954 (1,947)

18 Onions 254,099 252,711 (1,388)

19 Mushroom 41,782 40,873 (908)

20 Lettuce 173,904 173,469 (436)

21 Tomato 366,371 370,612 4,242

Product Category

Total Inventory in Supply Chain (lbs) • All 21 product categories, except Tomato, have less total inventory in supply chain of scenario 2

• Net saving in total inventory in

supply chain results from

o Inventory at retail stores will increase due to longer lead time

o Inventory at FFC will decrease in a larger amount due to risk pooling effect

May 22-23, 2013

MIT SCM ResearchFest 14

1. A Fulfillment Center provides benefits to Some Product Categories

The decision to add upstream produce facilities significantly depends on Product Categories and Locations which indicates

Demand Volatility and Supplier Replenishment Schedule

Supplier

Fulfillment Center

Grocery Distribution

Center

Retail Store

Channel 1

Channel 2

Product Category

Average demand

Standard deviation

ECV

BANANAS 1,423 238 0.17

MUSHROOM 44 11 0.26

BERRIES 15 30 1.97

WATERMELONS 57 62 1.10

Products for Channel 1:

Low Demand Volatility

Products for Channel 2:

High Demand Volatility

Conclusion

May 22-23, 2013

Channel Decision

MIT SCM ResearchFest 15

2. A Fulfillment Center adds Agility to the system

• Safety Time of supply chain and Total safety stock in the supply are reduced from Risk Pooling.

• However, a Fulfillment Center adds another “touch” to the system and may increase total time for all product categories

Conclusion

May 22-23, 2013

Q&A

MIT SCM ResearchFest 16 May 22-23, 2013