Poster

Client Context

Under Armour (UA) is an American sportswear manufacturer. Their largest distribution center, the Nashville Distribution House (NDH), is over 1 million square feet and processes 250,000-300,000 units a day. Not only does NDH serve as a distribution center, but also NDH handles processing returns, which are typically sent from wholesalers (e.g. Dick’s Sporting Goods). Returns items are part of Return Authorizations (RAs), which serve as a unique identifier for a return from a customer. Returned items arrive to the facility as pallets and are staged at the arrival docks until they are ready to be moved to the returns processing area. The operators at NDH do not know what is going to be on a pallet and do not know when a specific pallet is going to arrive; customers must send their items within 6 weeks of submitting an RA, and Under Armour waits no longer than 90 days for an RA to arrive. The returns process itself consists of the following steps: sorting, grading, reworking, and repackaging. After an item goes through the rework is processed, it is either returned to stock, sent to a client liquidation facility, or is donated/scrapped depending on its quality.

Project Objective

In 2018, UA began using a third-party reverse logistics company, Optoro, to complete their returns processing. After returns processing costs increased by 30% and returns processing times increased by 47%, UA is considering moving their returns processing to be back in-house, and they are starting with managing returns for their wholesale customers. NDH will become their hub for processing returns, and Under Armour wants to design a reverse logistics system that meets their goal of processing 20,000 to 30,000 returned units a day. Specifically, the UA Operations team is focusing on decreasing their returns processing cycle time, while also maximizing the profit obtained from returned items and minimizing labor/storage costs.

Two key opportunities were identified. First, UA could begin processing RA’s as soon as they arrive. Currently, UA waits for an RA to fully arrive before beginning to process any part of that RA, with a waiting limit of 90 days. Waiting for an RA to arrive increases staging time, which can cause products to lose value due to seasonality and can incur higher storage costs. UA also waits to process one RA at a time because they believe that pallets from the same RA are more likely to have similar overlapping SKU’s. However, provided data sets show that 52% of SKU’s in an RA have only 1-2 units; the second opportunity for improvement would be to process multiple RAs at the same time to further reduce cycle time while also overlapping the SKU’s across different RAs.

 

Design Strategy

Under Armour’s initial system design was a starting point for the team, but the problem was approached as a design problem; the focus was on designing a new returns process instead of building off of the initial system design. The design process was broken down into 3 main steps: 1) understanding system requirements, 2) creating a logical architecture to meet the system requirements, and 3) delivering an embodiment of our proposed logical architecture.

In order to understand the system requirements, the team identified functional requirements of the returns process after characterizing the inbound and outbound states of items. After solidifying the functional steps (sorting, grading, reworking, packaging), it was necessary to define a flow of the processes. After analyzing data, the team established the following flow: grading, sorting by product type, reworking with specialized reworkers, sorting by SKU, and finally packaging. When building the logical architecture, the team first researched returns operations and technologies before generating design options. Extensive analysis on historical returns data was conducted in order to determine the feasibility of each design option. Finally, a logical architecture was chosen based on the design feasibility that met the balance between our objectives.

To validate/test the approach, the team conducted data analysis on RA SKU overlap, SKU accumulation, put wall capacity constraints, put wall prioritization opportunities for downgrading quality, and gaylord usage. These analyses supported our design and affected our design’s final embodiment. A queuing model was also used to investigate cycle time, worker utilization, and throughput for our model.

 

Deliverables

First, the team provided implementations regarding new technology, new staging strategies, and a labor optimization model. The team recommends utilizing 2 new technologies: a timed put-to-light put wall and product type-specific gaylords. The put-to-light put wall will be used for Grade 1 items (93% returned items), and it will light up when a carton reaches capacity (determined by the physical appearance of carton “fullness”) or times out at 24 hours after the first item in the carton is placed. Using the put wall will decrease the time it takes to identify where to place each SKU into a carton and allow for a more organized putwall that is scalable and reconfigurable as needed based on SKU’s and product types. Product type-specific gaylords will be used for Grade 2 items, which are approximately 2% of returned items. Gaylords are used as a method of staging and allow for accumulation of Grade 2 items, which saves valuable putwall space by 50%, and gaylords allow for an increase in throughput of Grade 1 items in the putwall by 27%. Items in the gaylord will be processed when the gaylord reaches capacity, which will be between 3 to 30 days depending on the product type; instead of going through the putwall, the Grade 2 items will directly be sent to the packaging step, as they do not need to be reworked. The team recommends processing RAs as soon as they arrive and processing multiple RAs at the same time in order to reduce the cycle time associated with staging products. Beginning to process items as soon as they arrive immediately reduces the original average of 61 days of staging in NDH prior to processing. Processing multiple RAs at the same time will allow UA to combine SKUs across RAs and provide batching opportunities, which will save valuable space and reduce cycle time by not waiting to process one RA at a time. A labor optimization model was created in alignment with our final logical architecture, which will help UA distribute labor efficiently in the new returns process and reduce potential labor costs.

Second, a physical layout of the returns section of NDH was created to reflect the team’s final design embodiment; this provides direction for the UA operations team in creating a layout for their returns department. The various logical architectures are also detailed, with corresponding “efficiency” metrics (# of scans, touches, # of times in staging, and # of times to sort) and the number of operators for each step of the returns process.

 

Project Information

Fall 2019
Under Armour

Student Team

Veda Berg, Morissa Chen, Virginia Fishburn, Sarah Liu, Kyle Rasmussen, Farial Sufi, Rahul Verma, Kevin Zhou

Faculty Advisor

Faculty Evaluator