By Christian TeBordo

From bare grocery shelves in the pandemic’s early days, to giant freighters swarming off the West Coast, to the current baby formula crisis, supply chain disruptions have made headlines in the past few years. These issues are a result of a growing emphasis on efficiency on the one hand, and consumer convenience on the other. Many of the same processes that resulted in product shortages also enable next- and even same-day delivery. Striking a new balance is a global necessity.

At Rensselaer Polytechnic Institute, Jennifer Pazour, associate professor in the Department of Industrial and Systems Engineering, and Kaan Unnu, a doctoral graduate of the department, have proposed a promising step.

On-demand warehousing is an innovative model in which businesses store select inventory in third-party warehouses that have underutilized space, allowing them to fulfill orders quickly and efficiently.

Traditional distribution centers require big buildings, which are expensive to construct and maintain. This typically limits the number of distribution locations a given company can operate. Outsourcing all distribution needs to a third party can be time-consuming and may also come with minimum inventory requirements. In the right circumstances, on-demand warehousing can be a cost-effective solution.

Pazour and Unnu’s research offers the first optimization model incorporating on-demand system properties into distribution network design problems.

“On-demand warehousing is a new way for firms to create more agile, dynamic, and collaborative supply chain networks, which is especially beneficial if they need to adapt to uncertain and changing demand,” Pazour said. “However, such solutions also come with new cost structures and risks.”

On-demand warehousing doesn’t require the long-term commitment of constructing and maintaining a building, and it allows a company to spread inventory over more locations. But it also has drawbacks, like higher storage costs per unit.

To determine when on-demand warehousing is beneficial for customers, Pazour developed a dynamic model that simultaneously considers the location and allocation decisions of three types of distribution centers — self-distribution, third-party leasing, and on-demand. The research showed that several factors, including the availability of on-demand capacity, responsiveness requirements, and demand patterns, influence the cost-effectiveness.

“While on-demand warehousing was not found to be the best solution in every situation, it is very successful at supplementing more traditional distributions,” said Pazour. “We need to think differently about how supply chains acquire, utilize, and allocate resources to meet customer demands, and how we build the business models and algorithms to tap into underutilized resources.”

Hybrid network designs like those proposed by Pazour and Unnu’s model could be a key to navigating increasingly complex global systems.

“This research is exciting,” Pazour said. “It provides proven methods for firms interested in evaluating how best to adopt on-demand warehousing into their supply chain networks.”