YinJin Lee Freight Deliveries to Retail and Food...
1
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 25cm^3 50cm^3 1m^3 2m^3 4m^3 >4m^3 N_product/sum(N_product) Probability of delivery size by product type Accessories Apparel Apparel:Accessories Freight Deliveries to Retail and Food Establishments in Singapore Christopher Zegras, Professor, MIT-DUSP Edgar E. Blanco, Research Director, MIT-CTL • A city generates about 300 to 400 truck trips per 1000 people per day, and each person consumes about 30 to 50 tons of goods per year. Urban freight takes up 10-15 percent of the total miles traveled on city streets and 3-5 percent of urban land. • Urban traffic models mostly neglect freight activities due to lack of data. There is also lack of knowledge on how freight deliveries take place. • Current trip generation models are static and aggregated Not useful for policy evaluation. • Models based on store manager replenishment behavior can predict delivery frequencies and size, and supplier type. These variables can influence fleet size, vehicle size and origin. • Better freight behavioral models can improve traffic simulations under different regulation scenarios. Motivation / Background Observations Limitations and Conclusion YinJin Lee Ph.D. Candidate Engineering System Division Location choice Parking decision Establishment ordering behaviors Number of deliveries per stop Total number of stores Number of deliveries per store x ÷ Total trips attracted = Receiving establishments Walking route Driving route Inter-zone trip Stop/ Trip-end Delivery Tour (Multiple trips) Intra-zone trip Origin/ Source Data collection Establishment characteristics Weekly delivery frequency Total number of sources • Type of source • Product mix • Quantity • Delivery frequency • Time of delivery Establishment level Delivery level Interview surveys asked questions on two levels Hypothesis generated 1. Centralize distribution increase delivery size. 2. Centralize distribution do not affect delivery frequency or size. 3. Perishability the most powerful predictor of delivery frequency. 4. Except for congestion pricing there are little regulations pertaining to freight in Singapore, and we observe that deliveries mostly took place during off peak periods. • Delivery takes place when the operation cost is lowest. 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Another store Wholesaler Manufacturer Corporate warehouse Market share FashionApparelStore FoodCourt FoodService Pharmacy Food Stuff Multistage sampling First stage: Representative zones in the city. • Zones with The Most retail and food service outlets (SPH, 2013) in each cluster was selected. Second stage: Proportional sampling of establishments by industry types 671 surveys collected. There is a lot of variation in delivery frequencies that is not fully explained by industry types and product types. Econometrics models will be used to test hypothetical relationships between the variables. Food services and fashion establishments have high chance of being supplied by corporate warehouse. The chances are 34% and 71% respectively, compared to 8% for food courts. Perishability of food is a strong determining factor to delivery frequency. Delivery frequencies Prepared 5.01±2.76 Fresh food 4.40±2.23 Mix food 4.14±2.25 Frozen 3.39±2.42 Cosmetics 2.29±2.05 Dry food 2.28±2.09 Apparel-Accessories 2.26±2.61 Pharma 2.24±2.40 Apparel 2.11±1.77 Accessories 1.75±1.14 0 500 1000 1500 2000 2500 Time 0000 to 2500 (Next day) Legend ! Delivery before and after opening hours ! Delivery during opening hours ! Opening hours Cosmetics Fashion Food Court Food Service Foodstuff Pharmacy Most deliveries took place from 11am to 3pm. However food deliveries can take place before 7am, and dip between 11am to 1pm. Accessories are delivered in smaller packages while apparels are in larger packages. This can affect vehicle size.
Transcript of YinJin Lee Freight Deliveries to Retail and Food...
Research Expo 2014
0.00#
0.10#
0.20#
0.30#
0.40#
0.50#
0.60#
0.70#
0.80#
0.90#
1.00#
25cm^3# 50cm^3# 1m^3# 2m^3# 4m^3# >4m^3#
N_p
rodu
ct/sum
(N_p
rodu
ct)/
Probability/of/delivery/size/by/product/type/
Accessories#
Apparel#
Apparel:Accessories#
Freight Deliveries to Retail and Food Establishments in Singapore
Christopher Zegras, Professor, MIT-DUSP Edgar E. Blanco, Research Director, MIT-CTL
• A city generates about 300 to 400 truck trips per 1000 people per day, and each person consumes about 30 to 50 tons of goods per year. Urban freight takes up 10-15 percent of the total miles traveled on city streets and 3-5 percent of urban land.
• Urban traffic models mostly neglect freight activities due to lack of data. There is also lack of knowledge on how freight deliveries take place.
• Current trip generation models are static and aggregated à Not useful for policy evaluation.
• Models based on store manager replenishment behavior can predict delivery frequencies and size, and supplier type. These variables can influence fleet size, vehicle size and origin.
• Better freight behavioral models can improve traffic simulations under different regulation scenarios.
Motivation / Background Observations
Limitations and Conclusion
YinJin Lee Ph.D. Candidate Engineering System Division
Location choice !
Parking decision !
Establishment ordering behaviors!
Number of deliveries per stop !
Total number of stores !
Number of deliveries per store!
x!
÷!
Total trips attracted!= !
Receiving establishments Walking route Driving route
Inter-zone trip
Stop/ Trip-end Delivery Tour (Multiple trips)
Intra-zone trip Origin/ Source
Data collection
Establishment characteristics
Weekly delivery frequency
Total number of sources
• Type of source • Product mix • Quantity • Delivery frequency • Time of delivery
Establishment level
Delivery level
Interview surveys asked questions on two levels
Hypothesis generated 1. Centralize distribution increase delivery size. 2. Centralize distribution do not affect delivery frequency or
size. 3. Perishability the most powerful predictor of delivery
frequency. 4. Except for congestion pricing there are little regulations
pertaining to freight in Singapore, and we observe that deliveries mostly took place during off peak periods. • Delivery takes place when the operation cost is lowest.
0.00#
0.10#
0.20#
0.30#
0.40#
0.50#
0.60#
0.70#
0.80#
0.90#
1.00#
Another#store# Wholesaler# Manufacturer# Corporate#warehouse#
Market'sha
re'
FashionApparelStore#
FoodCourt#
FoodService#
Pharmacy#
Food#Stuff#
Multistage sampling First stage: Representative zones in the city. • Zones with The Most retail and food service
outlets (SPH, 2013) in each cluster was selected. Second stage: Proportional sampling of establishments by industry types 671 surveys collected.
There is a lot of variation in delivery frequencies that is not fully explained by industry types and product types. Econometrics models will be used to test hypothetical relationships between the variables.
Food services and fashion establishments have high chance of being supplied by corporate warehouse. The chances are 34% and 71% respectively, compared to 8% for food courts.
Perishability of food is a strong determining factor to delivery frequency.
Delivery frequencies Prepared 5.01±2.76 Fresh food 4.40±2.23 Mix food 4.14±2.25 Frozen 3.39±2.42 Cosmetics 2.29±2.05 Dry food 2.28±2.09 Apparel-Accessories 2.26±2.61 Pharma 2.24±2.40 Apparel 2.11±1.77 Accessories 1.75±1.14
0" 500" 1000" 1500" 2000" 2500"Pharmacy"Pharmacy"
FoodStuff"FoodStuff"
FoodService"FoodService"FoodService"FoodService"FoodService"FoodService"FoodService"FoodService"FoodService"FoodService"FoodService"
FoodCourt"FoodCourt"FoodCourt"
Fashion"Fashion"Fashion"Fashion"Fashion"Fashion"Fashion"Fashion"Fashion"Fashion"
Cosme:cs"Cosme:cs"
Time%0000%to%2500%(Next%day)%
Legend ! Delivery before
and after opening hours
! Delivery during opening hours
! Opening hours
Cosmetics
Fashion
Food Court
Food Service
Foodstuff
Pharmacy
Most deliveries took place from 11am to 3pm. However food deliveries can take place before 7am, and dip between 11am to 1pm.
Accessories are delivered in smaller packages while apparels are in larger packages. This can affect vehicle size.
