Warehouses are one of the most critical links in a supply chain. Storing and retrieving products to fulfil customers’ needs, wants and desires, your warehouse is the greatest tool you have to either satisfy, or totally disappoint, those who are eagerly waiting for their latest purchase.
So, it’s fair to say that your warehouse is a pretty big deal. But did you know that order picking is one of the most time-consuming activities to take place in a warehouse? It’s also one of the most expensive processes contributing to more than 55% of the total cost of warehouse operations.
Now that we have your attention, let’s take a look at some tools that will help you become an order-picking-efficiency-machine. In this post, we look at how intelligent algorithms can help you increase order picking activity while simultaneously reducing costs.
What is an algorithm?
Most of us would have ordered something online, and from the moment we hit “confirm order”, a series of human-less and automated processes take place to ensure that once the business receives the order, our purchase begins the many-fold process of order fulfilment.
Many of these processes are based on algorithms: processes and sets of rules to be followed by a software program. They can perform calculations, process data and automate reasoning tasks. Algorithms make it possible for software platforms to not only understand what to do, but how to do it.
There are all different kinds of algorithms that perform a range of different tasks from establishing what you see in your Google search results, to making it possible to securely share public keys between platforms and users. Today, we focus on looking at a very specific algorithm that is most suited to warehouse management.
How algorithms are used in warehouses
When it comes to order fulfilment, there is no doubt that efficiency is key. But if you have a warehouse packed with pickers that march out onto the warehouse floor every day to retrieve orders, you’re picking rate will only ever be as good as the amount of time it takes your pickers to receive the order, get to the product, and bring it back for despatch.
You might be thinking: “well I have the fastest pickers available!” - that may be so, however, in a warehouse that stores millions of SKUs, it’s impossible to assume your pickers can memorise, by heart, the location of each item. And that is where your productivity suffers: the time it takes your pickers to establish where items are, how far away from each other they are, and the toing and froing your pickers have to undergo due to inefficient routing.
An Amazon fulfilment centre in Hemel Hempstead seems to have locked down a pretty solid solution to this problem. You may be surprised to know, that pickers in this Amazon fulfilment centre don’t need to memorise a single SKU location across the whole 40,000sq m facility.
Workers are guided by software that calculates where the inventory is located and dispatches a human to retrieve it without the worker having to stop and think about anything. The facility's general manager, Henry Low, states that this method completely eradicates the need to learn where items are:
“The pickers are not meant to have to think too long about what they’re retrieving”. Thinking takes times, and the more thinking that takes places about where items are located, the more time it takes to fulfil your orders.
Amazon also puts a lot of thought into where they store each item in the warehouse. If you have a lot of items that look really similar, it could be incredibly difficult for pickers to differentiate between products. It is for this reason that Amazon warehouses appear as though they have been arranged by a seven-year-old, with Harry Potter DVDs located next to HDMI cables and so on. All of these measures are taken to reduce the amount of time it takes pickers to locate each individual item.
The software they use is able to work out the best possible route of getting to an item, while also avoiding any confusion or thought taken for a human to determine whether or not an item is the correct product. So what’s the magic behind Amazon’s software? Algorithms. Take a look below at how genetic algorithms are particularly well suited to warehouse-management.
Although it sounds as though Star Trek has met Planet Earth, genetic algorithms are a perfect fit for order picking because they are based on genetic principles in the structure of life but use techniques to find the most appropriate solution. For example, in the case of warehouse management a genetic algorithm would consider an order-picking route amongst 100s or 1000s of possibilities to find the best possible route.
The term genetic algorithm originates in computer science where a range of possible candidate solutions are tested against a potential problem to constantly evolve better solutions. They are often used to help see what would happen is something in biology, say a chromosome, mutated so we are able to better understand the impact changes would have on our genetics.
Genetic algorithms (GAs) are well suited when search techniques are required where the end goal is to find something faster. Optimising search is achieved by pooling together hundreds of possible solutions to find the one solution that offers the best possible way of operating.
When applied to the warehouse, it’s easy to see how GAs are a handy tool for operators to search for SKUs more efficiently. When applying a GA to your picking procedures, parameters and selection procedure play an important role in reaching the optimum outcome.
Developing an algorithm for better picking will help you establish better routing policies as well as the best possible sequence for picking. In a busy warehouse, it can be tempting to try and pick orders as quickly as possible by fulfilling orders as they come, however determining a better order and route for picking will save traveling time along aisles, which is actually a better solution for fast fulfilment.
Which layout is best for your warehouse?
We know from our previous post on the latest trends in warehouse management for 2018, that order batching is far more efficient than doing picks for single orders, but in order to gain the most productivity from order batching, it’s important to limit the amount of time spent picking. This is where algorithms can help you establish the perfect warehouse layout for your business.
Each business has its own unique warehouse layout, and it’s generally down to warehouse management systems to wear the burden of understanding every detail about goods and storage. New technologies, such as automated dispensing machines, can assist with the manual work involved with picking, however, it’s been shown that even with an array of new inventions made specifically for picking, the more traditional style of warehouse layouts are still the most popular.
Take a look at the image below, for example. The three examples of the traditional layout on the left haven’t been developed with efficient picking in mind. In fact, if your warehouse currently operates according to one of these layouts, it won’t really matter what technologies you’ve implemented to assist with picking because the structure of your warehouse hasn’t been optimised for the least amount of travel time.
Algorithms can be developed for optimal order-picking path design, and as such, could potentially change the way your warehouse is organised, or at the very least, change the route pickers take when retrieving items.
The image below shows a few different ways you can change the path taken when undertaking order-picking. If you take a look at the time saved between the S-Shape path compared to the Return path, it’s easy to see that the small amounts of time taken from making such a small change could mean the difference between being profitable or not-profitable in any given month.
We hope you enjoyed this post on how algorithms can change your warehouse layout. Let us know what you think about applying algorithms in the context of order-picking below.