The distribution of COVID-19 vaccines globally shone a light on cold chain logistics and temperature-controlled transportation, and therefore, the challenges the food industry faces. Vaccines made by Pfizer Inc. and Moderna Inc. are easily spoiled and must be kept at cold, specific temperatures, or they will be thrown away — just like food products.

With so many elements that affect carrier arrival times and the pressure to avoid hefty fines from retailers, there is an urgent need for data and analytics (D&A) systems to optimize temperature-controlled logistics, detect damaged goods, and sort effective storage layouts. Let’s dive in. 

Reducing Damaged Goods

Just like how 16,000 vaccine doses were potentially spoiled in Maine and Michigan due to temperature problems, temperature-controlled cold storage food deliveries can also be damaged and wasted. 

Physical damage can occur due to packaging integrity or mishandling, contamination, spoilage (where cargo is not ordered in a first-in-first-out approach), and temperature abuse (when food needed to be stored at zero degrees is kept at room temperature). These kinds of problems with supply chains are contributing to the 1.4 billion pounds of food wasted each year.

Computer vision systems can identify physical damage that can also happen as a result of temperature changes, by classifying images of products traveling through a distribution center and comparing them to non-damaged images. You can trace the damage depth and type and take an actionable approach to reduce further damage to your products.

Temperature abuse, leading to damaged goods, can also be related to external factors like transportation delays. Artificial intelligence (AI) and cloud data can solve this uncertainty by providing route optimization and ensuring that deliveries arrive on time at third-party logistics (3PLs) at a low cost and energy expenditure. 

AI continually retrieves data about roads and traffic, learns from it, and analyzes new methods to ensure drivers are taking the most optimized routes — in real time. The most efficient truck routes change on a near-daily basis, and constant route optimization ensures that companies are using all possible routes to avoid bottlenecks.

AI can also factor in historical dwell times at specific facilities, including congestion at inbound areas, and plan this into routes or suggest carriers to go to alternative distribution facilities.

Maintaining Efficient Cold Storage Layout

When customers depend on temperature requirements, warehouse managers need to have a deep understanding of where to place racks and how to optimize capacity in terms of customer inbounding. A cold storage facility is often separated into refrigerated units, areas for products that require blast freezing, and everything in between — and if you have a two million square feet facility, with 10,000 locations, these spaces should not be cluttered up with slow-moving freight. 

Goods can get spoiled during transportation but also due to delays in warehouse processes too. Effective warehouse slotting strategies are one way to approach space optimization and ensure that fast-moving freight and inflows of seasonal items are stored efficiently. This can improve productivity by 8 to 15%. 

One way of achieving this productivity boost is through a D&A solution called a Digital Twin, or a 3D digital replica of an entire warehouse. The digital twin architecture goes beyond just showcasing the model of objects; head office staff could monitor live data about the different categories of freight in real time through internet of things (IoT) sensors and camera feeds. They’d be able to see where to place inbound and outbound products in their facilities and work out the best path for warehouse staff to pick and store these items. Overall, this would help temperature-controlled shipments avoid unnecessary idling, particularly leading to fines in urban areas. 

Improving Staff Productivity

Remember the chronic shortage of workers and buckling supply chains during the run-up to Black Friday and Christmas in the UK in 2021?  Warehouses had to pay up to 30% more to recruit staff.

Other than energy, labor is the largest expense for cold storage distribution facilities, so warehouse managers want to get the most out of each man-hour, getting the right person to do the right task every minute of the day. But this is a massive challenge, especially when non-productive tasks like fixing broken pallets or searching for inventory get in the way and don’t add value to customers. 

Using D&A solutions helps identify the right employee for each type of task — both inbound and outbound. Machine-learning algorithms can be trained using case studies with a large number of warehouse datasets to target the storage allocation strategy (SAS) and the picking policy (PP). The outcome can mean that, if an employee picks a temperature-controlled pallet from inbound zone A to take to a storage rack, the ML can work out where the closest outbound pallet is, so return trips to the dock are optimized. 

Most traditional processes would have employees put a pallet in storage and go back to the inbound dock without picking up another palette. Task interleaving greatly reduces empty forks across distribution centers specializing in temperature-controlled logistics. 

This discovery helps to improve the resilience and organization of 3PL providers who need to assign incoming stock-keeping units (SKUs) to their storage systems. For example, United States Cold storage reported a 15% improvement in average turn times across warehouses, resulting in savings of over $300,000.

Adopting Cycle Counting Automation 

In a typical 3PL environment, about 40% of staff are dedicated to unloading or receiving freight. The size and complexity of customers can vary the workload dramatically; some customers send their cargo clearly marked and ready to pick up, others ship raw food items that require further processing, among other reasons.

A traditional 3PL still manually counts and verifies the cargo as multiple carriers arrive before putting the load into a storage facility, leaving room for human error when identifying products and quantity. But there’s an opportunity to save $147,000 per year in cycle counting labor.

Computer vision enhancements play a role in mitigating this problem. Drones can fly through aisles using computer vision to quickly automate the cycle count of inventory. The drones often don’t need pilots; engineers can draw a path for the GPS tracking drone in a warehouse and give clear instructions such as “don’t go higher than 28 feet.” They detect damage within the racks and alert warehouse managers or workers if they have stocked frozen broccoli instead of frozen peas, for example.

Monitoring Energy Consumption

Energy consumption is one of the largest expenses for 3PLs, especially when some goods require very precise temperature control. These types of facilities often face surcharges to purchase electricity during peak hours, meaning astronomical energy expenses. Think of Dallas, Texas, in mid-July: Everybody has their aircon on at noon. So, if a cold storage unit turns on its electricity too, it’ll face a surcharge. Fly-wheeling could greatly reduce energy expenses. 

Fly-wheeling relies on ML to find the best times to purchase energy from the grid, and AI can schedule decisions based on those predictions. If warehouse managers cooled their buildings way past the temperature requirement in advance, they wouldn’t need to consume during peak hours. This process effectively turns the facility into a battery. The company Lineage reported a 34% reduction in energy from using fly-wheeling methods over three years.

D&A can play an integral role in solving temperature-controlled logistic obstacles to drive safer, more efficient data-driven warehouse operations. But our recommendation is to always narrow down what the biggest challenge is in a facility instead of blindly deploying technology for the sake of it. Once you have identified a particular business problem, you can then pick the most appropriate technology, from simple descriptive analysis to complex AI algorithms.

Matthew Burgos is associate principal data consultant, and Ganes Kesari is co-founder, at Gramener.