In a Distribution Center (DC), walking time from one location to another during the picking route can account for 60% to 70% of the operator’s working time. Reducing this walking time is the most effective way to increase your DC overall productivity.
Scenario 1: Picking routes with 1 order picked per wave
I have published a series of articles that propose an approach to design a model to simulate the impact of several picking processes and routing methods to find optimal order picking by using the Single Picker Routing Problem (SPRP) for a two-dimensional warehouse model (axis-x, axis-y).
SPRP is a specific application of the general Traveling Salesman Problem (TSP) answering the question:
“Given a list of storage locations and the distances between each pair of locations, what is the shortest possible route that visits each storage location and returns to the depot ?”
This repo contains a ready-to-use Streamlit App designed for Logistics Engineers to test these different strategies by only uploading their own dataset of order line records.
- Improve Warehouse Productivity using Order Batching with Python - Article
- Improve Warehouse Productivity using Spatial Clustering with Python Scipy - Article
- Design Pathfinding Algorithm using Google AI to Improve Warehouse Productivity - Article
Based on your actual warehouse layout, storage locations are mapped with 2-D (x, y) coordinates that will be used to measure walking distance.
Warehouse Layout with 2D Coordinates
Every storage location must be linked to a Reference using Master Data. (For instance, reference #123129 is located in coordinate (xi, yi)). You can then associate every order line to a geographical location for picking.
Database Schema
Order lines can be extracted from your WMS Database. This table should be joined with the Master Data table to link every order line to a storage location and coordinate its (x, y) in your warehouse. Extra tables can be added to include more parameters in your model like (Destination, Delivery lead time, Special Packing, ..).
For more information and details about calculation: Medium Article
For this study, we will use the example of E-Commerce type DC where items are stored in 4 level shelves. These shelves are organized in multiple rows (Row#: 1 … n) and aisles (Aisle#: A1 … A_n).
Different routes between two storage locations in the warehouse
- Items Dimensions: Small and light dimensions items
- Picking Cart: lightweight picking cart with a capacity of 10 orders
- Picking Route: Picking Route starts and ends at the same location
Scenario 1, the worst in terms of productivity, can be easily optimized because of
- Locations: Orders #1 and #2 have common picking locations
- Zones: orders have picking locations in a common zone
- Single-line Orders: items_picked/walking_distance efficiency is very low
Scenario 2: Wave Picking applied to Scenario 1
The first intuitive way to optimize this process is to combine these three orders in one picking route — this strategy is commonly called Wave Picking.
We are going to build a model to simulate the impact of several Wave Picking strategies in the total walking distance for a specific set of orders to prepare.
In the article, I have built a set of functions needed to run different scenarios and simulate the picker's walking distance.
Function: Calculate the distance between two picking locations
Function: Different routes between two storage locations in the warehouse
This function will be used to calculate the walking distance from points i (xi, yi) and j (xj, yj).
Objective: return the shortest walking distance between the two potential routes from point i to point j.
Parameters
- y_low: lowest point of your alley (y-axis)
- y_high: highest point of your alley (y-axis)
Function: The Next Closest Location
Function: Next Storage Location Scenario
This function will choose the next location among several candidates to continue your picking route.
Objective: return the closest location as the best candidate
This function will create your picking route from a set of orders to prepare.
- Input: a list of (x, y) locations based on items to be picked for this route
- Output: an ordered sequence of locations covered and total walking distance
Function: Create batches of n orders to be picked at the same time
- Input: order lines data frame (df_orderlines), number of orders per wave (orders_number)
- Output: data frame mapped with wave number (Column: WaveID), the total number of waves (waves_number)
Function: listing picking locations of wave_ID picking route
- Input: order lines data frame (df_orderlines) and wave number (waveID)
- Output: list of locations i(xi, yi) included in your picking route
After setting up all necessary functions to measure picking distance, we can now test our picking route strategy with picking order lines.
Here, we first decided to start with a very simple approach
- Orders Waves: orders are grouped by chronological order of receiving time from OMS ( TimeStamp)
- Picking Route: The picking route strategy follows the Next Closest Location logic
To estimate the impact of wave picking strategy on your productivity, we will run several simulations with a gradual number of orders per wave:
- Measure Total Walking Distance: how much walking distance is reduced when the number of orders per route is increased?
- Record Picking Route per Wave: recording the sequence of locations per route for further analysis
Experiment 1: Results for 5,000 order lines with a ratio from 1 to 9 orders per route
_For more information and details about calculation: Article
Order Lines Processing for Order Wave Picking using Clustering by Picking Location
Group picking locations by clusters to reduce the walking distance for each picking route. (Example: the maximum walking distance between two locations is <15 m)
Spatial clustering is the task of grouping together a set of points in a way that objects in the same cluster are more similar to each other than to objects in other clusters.
For this part we will split the orders in two categories:
- Mono-line orders: they can be associated to a unique picking locations
- Multi-line orders: that are associated with several picking locations
Left [Clustering using Walking Distance] / Right [Clustering using Euclidian Distance]
Grouping orders in cluster within n meters of walking distance
Example: Centroid of three Picking Locations
Grouping multi-line orders in cluster (using centroids of picking locations) within n meters of walking distance
To sum up, our model construction, see the chart below, we have several steps before Picking Routes Creation using Wave Processing.
At each step, we have a collection of parameters that can be tuned to improve performance:
Methodology: Model Construction with Parameters
Methodology: Three Methods for Wave Processing
We’ll start first by assessing the impact of Order Wave processing by clusters of picking locations on total walking distance.
We’ll be testing three different methods:
- Method 1: we do not apply clustering (i.e Initial Scenario)
- Method 2: we apply clustering on single-line orders only
- Method 3: we apply clustering to single-line orders and centroids of multiline orders
- Order lines: 20,000 Lines
- Distance Threshold: Maximum distance between two picking locations (distance_threshold = 35 m)
- Orders per Wave: orders_number in [1, 9]
Test 1: 20,000 Order Lines / 35 m distance Threshold
- Best Performance: Method 3 for 9 orders/Wave with 83% reduction of walking distance
- Method 2 vs. Method 1: Clustering for mono-line orders reduce the walking distance by 34%
- Method 3 vs. Method 2: Clustering for mono-line orders reduce the walking distance by 10%
Because the resources provided by Streamlit Cloud or Heroku are limited, I suggest running this application locally.
sudo pip3 install virtualenv
virtualenv venv
source venv/bin/activate
pip install -r requirements.txt
streamlit run app.py --server.address 0.0.0.0
Instructions: Click on the URL
-> Enjoy!
This app has not been deployed, you need to use it locally/.
This Streamlit Web Application has been designed for Supply Chain Engineers to help them simulate the impact on picking route optimization on the total distance of their picking operators.
- You can use the dataset located in the folder In/df_lines.csv
- You can build your own dataset following the step of ('Initial Step') above
Experiment 1: Parameters
As the computation time can increase exponentially with the size of the dataset (optimization can be done) you can ask the model to take only the n thousands first lines for analysis.
In the picture below we ask the model to run a loop testing scenarios with the number of orders per wave varying between 1 to 10
Click the button to start the calculations
Experiment 1: Final Resultss
💡 This is the same graph with the one presented in the article
Experiment 2: Parameters
As the computation time can increase exponentially with the size of the dataset (optimization can be done) you can ask the model to take only the n thousands first lines for analysis.
Click the button to start the calculations
Experiment 2: Final Results
💡 This is the same graph with the one presented in the article
Senior Supply Chain and Data Science consultant with international experience working on Logistics and Transportation operations.
For consulting or advising on analytics and sustainable supply chain transformation, feel free to contact me via Logigreen Consulting \
Please have a look at my personal blog: Personal Website