Order Processing Systems Capacity – A Key Design Specification for Distribution Automation and Peak Season Shipping That Should NOT Be Overlooked

Your order processing system’s capacity for peak season shipping periods might have fatal flaws. You need to focus on key design specifications for distribution automation that simply CAN NOT and SHOULD NOT be overlooked!

When companies automate distribution facilities, there are many success stories (you see lots of them on LinkedIn or other social sites). However, there are many not-so-successful implementations you don’t hear about, or ones that weren’t completely off the rails but never came close to achieving the performance or financial goals that they were sold on.

There are many reasons for these less-than-stellar implementations: legacy systems that can’t keep up with islands of automation dropped in, mismatches between the inventory and order models and the automation that was implemented, software systems that can’t control a high-speed/real-time environment, etc. But sometimes, it’s something much more rudimentary: a lack of understanding of, and planning for, what’s actually required to process orders in these systems, within a constrained timeframe. In other words, the actual true pipeline “burst” capabilities of an order processing system.

Example question: What are the actual throughput requirements and time constraints that a system has to process within 1-2 hours during a peak season shipping on a peak day(s) with hard wall time limits to tender packages to carriers for shipment? This key metric may be the true design criteria, especially in a world of same-day/next-day shipments and later inbound sales order processing cut-off times. How much does a pipeline burst metric change the underlying design concepts and requirements, and what are the true design goals?

To illustrate this, let’s look at a real company that, after doing their own analysis and developing an automated system concept with an automation product manufacturer, had Integrated Systems Design (ISD) perform a complete inventory/order analysis and forecast, along with possible options for system concept designs. Here is a very small snippet from the analysis that deals specifically with pipeline burst metrics.

Let’s start with the baseline operational requirements this company set forth and the basis of their initial automated design:

• Core distributor of consumer goods / products
• Orders placed by 5 PM are shipped same day (promoted / guaranteed CSA _customer service level agreement)
• Order & line growth was forecasted at 10% year over year over 5 years
• UPS/FedEX ground shipments needed to be tendered by 6:00 PM for truck pulls in this location to maintain time in transit targets for the carriers and hit their master sorting schedule.

Pretty normal stuff. Below are the baseline numbers, with the 10% forecasted year-over-year growth factored in. However, the orders/lines per hour were not in the original data, which is a critical omission. Integrated Systems Design (ISD) added these later

Nothing crazy here either. A normal forecast. The daily data was originally used to the size a system from a throughput perspective. The initial design target was set for year 3, with the initial system implementation expanding in year 4. Obviously, this high-level analysis does not include individual item cube or velocity calculations, just the transactional requirements the picking, packing, replenishment, manifesting, etc., will need to contend with.

However, as we start working through some different elements that will impact these numbers, the underlying assumption of lines/orders per hour began to change dramatically.

This company, like many others, has seasonal/holiday increases in volumes. It starts in October and runs through the end of December. The original forecast numbers above were based on a YEARLY average. Analyzing the true order profile, about 44% of all outbound order / line volume is processed within these three holiday months, which is common in many industries. Here are the updated requirements based on that impact:

As you can see, the numbers jump significantly. Over the past three months, the hourly order rate has nearly doubled (we will continue to monitor this metric as we progress). In fact, year 1 is now greater than year 5 of their initial forecast during this timeframe. Instead of 267 orders per hour, the operation must process an average of 500 orders per hour, in the exact same timeframe. Based on the original system concept that was supposed to absorb order volume until year 3 (then be expanded), these results show that the processing requirements surpass the initial system concept requirements in year 1 during this critical timeframe.

However, there are more impacts to the model. Like many companies, the inbound order volumes during the week are NOT consistent. Mondays and Wednesdays were 60% of the weekly order volume, with Mondays accounting for 38% of total order/line transactions. Again, this starts to move the order volume away from averages to the true processing requirements that automation must contend with, and why the pipeline calculation becomes so critical.

Here’s what this data looks like on a typical Monday during that 3-month peak season shipping:

Now things are getting even more dire. During the peak season shipping period, on an average Monday, the year 1 requirements are 950 orders at 3,326 lines per hour, ballooning to 1,391 orders and 4,869 lines per hour in year 5. This is 3.5 times the original design specification. The mean original system concept would have failed under this scenario immediately in year 1.

But there’s even more. As with many companies with a core online/ecomm presence (or salesforces that place orders from the road), when orders come into the system, they play a significant role in order availability and process/time constraints in the physical operation. Many times, these order groups will be consolidated within certain timeframes. The requirements then become much more focused on the actual hourly rate (Pipeline Burst), as the system design MUST take into account inbound customer orders WITHIN the UPS/FEDEX deadlines of the 6:00 cutoff/truck pull. The daily rate blended average becomes somewhat secondary.

In this particular case, the company received 53% of inbound customer orders between 3-5PM, qualifying for same-day shipment based on the published CSA’s. Remember the cutoff for tendering orders to the carriers has a hard wall limit of 6PM.

If we look at the data with this applied to the peak season, on a Monday, with 53% of the orders coming in after 3PM for same-day shipping, these are the actual requirements in terms of moving orders through the system in a constrained timeline:

The required hourly rate is nearly FIVE TIMES the initial study and the core system design they were considering. Here’s a direct comparison between the original average forecast and the pipeline forecast:

These analysis results clearly show how much capacity the original concept was designed for and how much transactional firepower is required. If this system had been implemented in year 1, during the critical holiday season, on a Monday, the system would have fallen short by 667 orders/3,464 lines every hour in the last 3 hours of the day, meaning 2,000 orders would have missed the CSA targets – every Monday. The impact that shortfall would have had on their customer base would have been devastating.

Working overtime/extended shifts does not help, as the company can finish processing orders internally, but they won’t make the UPS truck, and a later truck from UPS won’t make the required sort time at the UPS/FedEX facility. This means that the CSA commitments won’t be met (with a financial penalty of free shipping in this case). As a reminder, this does not only affect the Monday’s, as the entire peak season volume would swamp the original system concept.

Step back and think of the very real impacts on the ENTIRE operation under these scenarios. It’s not just the core picking automation/process, but all the functions in and around it:

• Orders need to be allocated and sent to the floor as quickly as possible – which can be a struggle for some WMS under heavy order loads.
• Those orders need to be physically inducted into the automation and flow in and out of the pick stations at the required rates (tote/carton selection, induction, labeling, routing, etc.).
• Pickers need to pick at the rates required, sustained for that timeframe. For example, if a picker can process 350 lines per hour out of the automation and the system is based on the original forecasted numbers, that would be 3 pickers. During these peak season shipping times, there would need to be 10-12 pickers. That requires much more firepower in terms of sheer order and product routing and delivery. It requires many more pick stations, much more complex routing, both order and product delivery management, and more labor content.
• Even in a robotic system where additional robots can be added temporarily, can the current labor and human/system interfaces keep up, or do those need the ability to flex for additional resources (both human and system)?
• Also, there will be an increase in hot replenishments required for orders in process, as those orders can’t be picked because there is not enough inventory in the primary pick locations (also slowing the pick operation down even more). Static min/max’s may not be able to react quickly enough to accommodate these spikes, so replenishments become much more frequent.
• How does newly receipted product get loaded into a system that is already at its maximum transactional capacity?
• Has the system been designed or can it be ramped up in terms of core inventory storage capability in terms of daily order processing? Days of inventory mean more storage when the overall units per day increase this dramatically.
• How does this mass of orders flowing through the packaging areas affect that operation, and in terms of a potential bottleneck, possibly negatively affecting the picking processes as well? If a person can package an order every 2 minutes and the system was geared for 243 cartons per hour, it would require 8 packagers, but at 910 per hour, the system would require 30 packagers. How will all those orders be routed, to what physical stations, and to what labor pool? Are there ways to streamline this critical operation?
• Can manifesting, carton labeling, document printing, and order staging work at these increased rates – or will they cripple the packing areas, which will cripple the picking process? Has the system been designed to handle that pipeline requirement?
• What about other elements, like loading trucks, sorting orders, and case/pallet/specialty picks that may be processed elsewhere and may have to be coordinated on the docks?

If any of these systems fall behind, the order processing will FAIL to meet the CSA requirements. Based on the original data set and original concept designed around it, that level of failure was engineered into the original concept system. This is also why islands of automation, not properly integrated into the processes and systems around them, can spell potential disaster.

It is critically important to understand the entire process requirement matrix of an operation, from initial product receipt to orders out the door and everything in between, in order to properly design an operation and the equipment/automation that is the best fit to run it. These types of analyses, done correctly, are paramount to adopting any kind of automation.

If you need help understanding your requirements, inventory & order profiles, forecasts of growth, labor requirements & costs, and an UNBIASED view of system concepts and what works best for your operation, with a company that will implement these systems and support you through the entire design, installation, and commissioning process and beyond, please feel free to reach out to me for a free consultation. Robert Jones Integrated Systems Design (ISD)