Authors: Natalie Feng, Eric Wu

1. Final Project:
Ultrafast Grocery Delivery
Natalie Feng, Eric Wu
Dec. 21, 2021

2. Traditional Grocery Store Existing Delivery Service Fast Delivery Service
Mature: order 1 day beforehand, 1-2h delivery window -> New: 15-20 min fast delivery
Grocery Shopping Experience

3. Ultrafast Grocery Delivery
● Faster delivery Service
● Smaller store size
● More locations
● Less SKUs (1000-2000s
compare to 40000s in typical)
● Lower start cost
● Quick expansion
(suburban)
Grocery Shopping Experience
Source: Ultrafast delivery: The $28B market to build the on-demand bodega. (n.d.). https://sacra.com/research/ultrafast-delivery-online-grocery-market/

4. Customer Perspective
● Place order at least 3 hours in advance
● Select an available delivery window
● Stay at home wait for the delivery
Operator Perspective
● Gather all orders in the same delivery window
● Solve the Vehicle Routing Problem
● Assign to delivery drivers
(Amazon app ordering
page)
Benchmark System - Whole Foods Market

5. Customer Perspective
● Place order and get an ETA
● Pickup order in 15-30 mins
Operator Perspective
● Pass the order to shopper, pack the order; Assign the order to a delivery driver
● Shopper finish packing, hand the order to delivery driver (about 3 mins)
● Delivery driver hit the road for delivery
Alternative Design - Ultrafast Delivery

6. Data Collection: Set Study Area
Manhattan tracts vs Upper East Side Population distribution in Upper East Side tracts

7. Methodology
Benchmark System Alternative System
WholeFoods Stores 15 mins & 30 mins Ultrafast Delivery
Capacitated Vehicle Routing Problem Location Allocation - Set Covering Problem
● Location of Whole Foods Stores
● Demand Simulation - 5W/2W
● Vehicle Routing - Optimal
Distance
● Cost Analysis - Cost Function
● Weight-based Location Allocation
● Demand Simulation - 5W/2W
● Vehicle Routing - Optimal Distance
● Cost Analysis - Cost Function

8. Warehouse locations for 15-minute and 30-minute delivery service
LSCP: Find Warehouses for Alternative Systems
Tool: Python Package: spopt

9. VRP_Generate Random Demand
Orders in each time-slot Weekday Weekend
Situation 1
(140 orders/day)
Low 0-15 3 2
Medium 15-35 3 2
High 35-50 1 3
Situation 2
(450 orders/day)
Low 20-50 3 2
Medium 50-80 3 2
High 80-120 1 3
Demand Assumptions: 7 time-slots in a day, 2 hours per time-slot
● In Sacra’s report, it suggest a mature store should have a daily order of 500.

10. VRP_Run VRP for Each System
Situation 1 (mi) Mon Tue Wed Thu Fri Sat Sun Total
Whole Foods 29.64 29.25 29.61 31.95 30.09 34.38 35.72 220.64
30 min 108.91 102.93 109.15 105.88 112.03 136.06 151.45 826.38
15 min 168.02 154.21 170.68 163.33 168.84 227.16 233.54 1285.76
Situation 2 (mi) Mon Tue Wed Thu Fri Sat Sun Total
Whole Foods 65.97 61.60 62.71 65.10 61.38 49.01 49.90 415.66
30 min 230.06 228.37 220.99 210.17 227.98 259.01 255.63 1632.19
15 min 384.61 368.96 393.09 349.65 373.93 440.18 431.66 2742.07
Tool: Python Package: ortools
Average mileages from 3 random demand sets.

11. Cost Analysis: Evaluate Each System
Alternative-15mins Alternative-30mins Benchmark-Wholefoods
General Warehouse Rent: $150/yr/sqft; Bicycle Speed: 7.5 mph
Processing time 5 mins 5 mins Not important
Travel time 10 mins 25 mins 2 hrs
Maximal coverage 1.25 mi 3.125 mi everywhere
Deliver capacity 1 order/deliver 1 order/deliver 15 orders/deliver
Deliver wage $30 per hr $30 per hr $40 per hr
Delivery cost $4 per mile $4 per mile $5.3 per mile
Cost Assumptions

12. Cost Analysis
Average Mileage (mi) Unit Cost ($/mi) Delivery Cost Fixed Cost Total
Whole Foods 294.19 5.3 $ 1,559.21 $ 25,000 $ 26,559
30 min 1101.84 4 $ 4,407.36 $ 9,375 $ 13,782
15 min 1714.34 4 $ 6,857.36 $ 12,500 $ 19,357
Situation 1: 3 sets of 7-day demand, computed average values
5.83x
3.75x

13. Average Mileage (mi) Unit Cost ($/mi) Delivery Cost Fixed Cost Total
Whole Foods 554.21 5.3 $ 2,937.33 $ 25,000 $ 27,937
30 min 2176.25 4 $ 8,704.99 $ 9,375 $ 18,080
15 min 3656.09 4 $ 14,624.35 $ 12,500 $ 27,124
Situation 2: 3 sets of 7-day demand, computed average values
6.6x
3.93x
● More difference in mileage, Higher delivery cost.
● Increasing in order size leads to less profit margin.
Cost Analysis

14. Discussion
● Assumptions on cost play a big role
● More travel mileage for a faster service
● Customers’ willingness to pay for a premium
● Stay competitive with the booming market