Supply chains are a critical, and often unnoticed, part of our everyday lives. Almost everything that we purchase in a store comes to us as a part of a supply chain and managing these networks is a complex and ever evolving task.
Location intelligence plays a critical role in supply chain network design.
From finding better ways to serve stores, the optimal location for a new distribution center, or understanding how goods are reaching their final destination, every aspect of supply chain design is tied to location data.
In this post we will show you the impact and importance of a well-designed and optimized supply chain for any business.
Supply chain network design and location intelligence
Supply chain network design is the process of building and modeling a supply chain to better understand the costs and time associated with bringing goods to market with the resources and locations available.
Some questions that are commonly evaluated as a part of this process are:
- How do I design my supply chain network to deliver the required service at the lowest possible cost?
- Given a fixed network, how do I determine optimal product sourcing and inventory deployment rules to meet anticipated customer demand?
- Given a logistics network and a defined distribution strategy, how can I best use my available transportation resources?
The end goal is to create the most efficient network possible, meet the demand of customers, and ensure the lowest possible cost to serve your network.
This process includes many different variables and models but many of them are tied to location, such as your distribution centers, store network, and possible routes to serve those stores.
Other assumptions, such as number of transportation resources, assumed delivery time, and total route time are also tied to location even though they might not initially appear to be impacted by location. The exact routes and road networks play a major role in how you will ultimately design your routes and assign resources to different clusters of stores.
Evaluating the Publix supply chain network
To show how location intelligence can impact supply chain network design, we analyzed the supply chain network for Publix, a leading grocery store chain in the southeastern United States, with primary operations in Florida.
Publix has over 1,110 stores served by 8 distributions centers.
The company expanded into North Carolina in 2011 and Virginia in 2016 and will likely continue to open new stores in these states.
However, the company has not opened a new distribution center to serve these new locations.
We evaluated Publix store and distribution center locations to understand how Publix is currently serving their stores, where Publix should place a new distribution center, and to quantify the ROI of opening a new distribution center. To do this we used the following data sources and analyses:
- Publix store locations and distribution center locations
- Demographic data from the Data Observatory to see the population served by each store
- Spatial clustering analysis to create logical clusters of stores for each route
- Optimized routing, or the most efficient route from a start/stop location to a set of other locations, to see time and distance of each route
To understand the current supply chain network design we:
- Created logical clusters of stores using the clustering analysis in CARTO to find logical groups of stores.
- Next, we assigned these groups to the nearest distribution center, and made some minor adjustments for outliers that needed to be assigned to different distribution centers.
- We used the same clustering analysis to create logical clusters for each of the routes originating from the distribution center.
- We used the optimized routing to find the most efficient route from the distribution center, to each of the stores, then back to the distribution center.
This analysis gives us the length of the trip and the drive time for the entire route. After looking at the data, some routes needed to be modified to make sure they could be completed within a standard shift, or if that was not possible, that route would need to be split into two shifts.
Assuming that a drop off takes 30 minutes, we then optimized the routes even further and split some of the routes into smaller routes to ensure we had as many routes as possible that could be completed in one shift.
In the resulting map you can see that to use each route once, it would take:
- 41,554 miles
- 1,134 hours in driving and delivery time
- 101 total routes
You can see that the Florida routes are shorter and well optimized apart from a few routes, but the Dacula, GA (purple circles) and McCalla, AL (light green circles) distribution center routes are very long and serve a significant amount of stores.
Dacula: 37% of Total Distance, 26% of Total Time, 24 Routes
McCalla: 22% of Total Distance, 14% of Total Time, 12 Routes
Using CARTO we can see some obvious improvements such as McCalla 5 (the green route to Chattanooga), which is better served by the Dacula 6 route (the purple route to Knoxville).
Using the final visualization, we can see that we need to add a distribution center to better serve the stores furthest away from the Dacula, GA distribution center since it serves almost 2 times more stores than any other distribution center.
Adding a new distribution center
After looking at the stores that the Dacula distribution center serves, we will want to place a new distribution center somewhere in North Carolina to serve these stores and to allow for more expansion in North Carolina, South Carolina, and Virginia.
To select this location we looked at data from the Data Observatory, specifically the employment, total population, and roads data to find a large city near major highways with connections around the region.
We narrowed this down to three candidates:
All cities are suitable based on the population and employable population. However, after looking at the proximity to the current store footprint and proximity to major roads, it is clear that Charlotte is the best suited location for the new Distribution Center.
Outcomes of the network optimization
The process for analyzing the new routes is almost identical to the process above: create logical clusters, analyze the routes, review and refine the routes.
After running this same analysis with the new store clusters and distribution center (resulting map here and below), we noticed some significant optimizations in the supply chain network from our original map at the top of the post:
- 15.7% decrease in distance driven
- 6.4% decrease in overall delivery time
- 9 fewer routes
This all accounts for estimated annual savings of $950,000 to $1.1M savings in fuel and time costs.
There are other additional benefits which are not accounted for, such as increased ability to serve stores and deliver fresh foods.
Overall it is clear that adding a new distribution center would not only save costs in the near term, but would also enable Publix to add somewhere around 80 additional stores in the North Carolina, South Carolina, and Virginia markets since the new Charlotte distribution center is currently only serving 81 locations.
Why CARTO for Supply Chain Network Design?
You can conduct supply chain network design with any location intelligence tool, but we’re a little biased here.
As you can see in the examples above, CARTO has several key advantages that allows it to run complex Supply Chain Network Optimization analyses like the Publix use case:
- Bring your data to the platform and analyze it with ease
- Integrate with other data such as demographics and advanced optimized routing
- Visualize the optimized network and routes on a map, not just the data
CARTO brings together the data, analysis, and visualization to make complex analyses like supply chain network design and optimization possible and allows you to not only understand, but visualize and act on the insights derived from your analysis.
Have a supply chain you’re designing? Request a demo and we’ll walk you through the same process with your own data.