Poster
Client Context
Children’s Healthcare of Atlanta (CHOA) operates as a not-for-profit organization well-renowned for their pediatric care to children in the state of Georgia. The Children’s system is comprised of 3 hospitals (Egleston, Scottish Rite, Hughes Spalding), 8 urgent care facilities, 58 neighborhood clinics, and 11 telemedicine sites in Georgia.
CHOA’s warehouse at 375 DeKalb Industrial Way in Decatur (DIW) serves as the primary focus of our project. This 21,420 square foot warehouse is a just-in-time facility servicing all 53 of their metro-Atlanta locations, comprised of 918 requesting locations representing different functional areas within the hospitals and clinics.
The warehouse is a 24/7 operation that employs 25 employees across 3 shifts. At inbound, bulk shipments of items are received from suppliers. These bulk shipments are broken down into item boxes or individual pieces and then stored either in small blue bins on the storage shelves, directly on the shelves, or on pallets. Once orders are received from clinics or hospitals, they are printed on physical pick tickets organized alphabetically by each item’s pick location. The workers operate within the approximately 10,800 square foot picking area to pick the items for each order into colored bins and place the completed orders onto carts which are shipped to the various facilities. DIW works with 1,669 SKUs and fulfills approximately 128 outbound orders per day (1,277 per week).
Since DIW is CHOA’s only warehouse, there has been a need for additional storage space in the past year. Therefore, they have been using an area attached to one of their clinics in Gwinnett as a small auxiliary storage facility for overflow inventory. This facility primarily holds excess personal protective equipment (PPE) from the ongoing COVID-19 pandemic.
Project Objective
In 2025, CHOA plans to open a new hospital located at their new North Druid Hills campus. This hospital will replace CHOA’s current Egleston hospital and will bring a 116-bed increase to CHOA’s system. While it was known that additional beds would increase demand on DIW, it was unclear at the start of this project exactly how this growth would affect warehouse operations, or if DIW would be able to handle the increased demand. Therefore, our goal for this project was to help CHOA more clearly evaluate how this growth would affect DIW, assess if DIW would be able to meet this demand, and finally provide CHOA with recommendations on how to prepare for this future growth, specifically answering the question of whether an investment in a new facility would be needed.
Design Strategy
To analyze the capacity of the warehouse, we focused on the largest room in the warehouse, which is where all the shelving units used for picking are. This area is about 10,800 sq ft and houses about 1700 SKUs. We conducted a space analysis of the warehouse where we noted the amount of space each item is currently allocated in the warehouse as well as the shelving capacity of each shelving unit. Currently the warehouse is using 90% of its maximum shelving space, which we will reference as space utilization
In order to assess the utilization of DIW’s workforce, we built a Simio model that shows the order picking process. We conducted time studies at the warehouse to collect data on the picking process. We then fit continuous empirical distributions to this data for all of the process time distributions within the simulation. We ran the simulation using one month of real hospital and clinic orders, and used pre-COVID data, so that our results would not be affected by pandemic changes. After running our current state model, we found that the employee utilization is 89% and the average order pick time is 10.54 minutes.
Once we analyzed the current capacity and worker utilization, we then looked to quantifying CHOA’s future demand growth to assess the effect on warehouse operations.
Using historical transaction data from 2015-2020, we wanted to build regression models to predict transaction quantity, also known as number of items, requested from the warehouse by CHOA’s 918 requesting locations. Once all error assumptions were tested and the models were validated with 2020 data, we could move into using the equations to forecast for the years 2021-2025. We used a linear or exponential trend-line equation to predict the changes in the number of requesting location and transaction count variables depending on the historical trends of these variables. Overall, our forecast predicts a 22% increase in CHOA system demand.
We then applied this forecasted growth to our current state to see how it would effect the system. Over the next four years, we predict that the space utilization will go from the current state of 90% to 110%. Suggesting there will be a shortage of roughly 1,200 square feet.
In order to assess the effects of demand growth on worker utilization, we used the demand forecast model to increase the number of orders in the simulation run. After running this scaled-up version of the Simio model, we found that the employee utilization increased to 97%, an 8% increase.
Since our analysis suggested DIW would not be able to handle the increased demand with their current operations, our next step was to evaluate if changes could be made to DIW which would allow them to handle the future growth. Our team identified various opportunities within the warehouse to improve space utilization and reduce worker utilization. As part of our efforts to improve space utilization in DIW, we first focused on updating CHOA’s par values. For CHOA, a Par Value is the minimum quantity of inventory needed while also providing a cushion for delivery lead times. CHOA’s current par values were not updated for over 10 years and were causing CHOA to over-order and over-store many items. CHOA’s ideal Par Values are 2 weeks of supply + safety stock. By using a centered moving average model, we forecasted the monthly item usage of all the 1700 SKUs. From this monthly forecast, we got the weekly item usage and used the lead times provided to us by CHOA. After calculating new par values, we then compared our par values to CHOA’s par values and found that our results had a 34% increase in accuracy.
Another opportunity to improve space utilization was Stagnant Item and Overstock Identification. Before, it was hard to assess what items could be reduced and by how much. With our Updated Par Values, we were able to assess the current overstock with quantitative data. Items with a Stock on Hand more than 100% above the par value should be flagged for evaluation. From the active 1650 inventory items, 823 items (or around 50%) were flagged as overstocked. To allow CHOA to easily evaluate stagnant and overstocked items on a regular basis we created an excel macro.
In an attempt to try to decrease CHOA’s worker utilization, we decided to create an optimization model to optimally place the items throughout the warehouse. We then grouped SKUs into 438 subgroups and used them in the Bin Optimization Model, which is a mixed integer linear program based off the Knapsack problem that assigns groups to shelving units.
The objective of the model is to minimize the distance to the staging area based on priority of the frequency a group in orders. We then input the new item locations into the Simio model and found that the employee utilization during the picking process decreased by 3%. This decrease proves that the optimization model was effective in reducing picks times but does not capture the additional value of the reorganization model which also optimizes the restocking process as this process is not simulated in the Simio model.
Through our demand forecast, we have predicted a 22% increase in system demand by 2025. Even with the potential 13% reduction in space utilization, our analysis has showed that DIW will not be able to handle the predicted increase in demand. Therefore, we believe that CHOA will need to expand their operations to a new warehouse, while still implementing our recommended labor and inventory strategies.
Since our analysis led to the conclusion that Children’s will need a new warehouse to manage the increase in demand, we wanted to give further recommendations about the size and placement of this new facility. We conducted an analysis of the amount of additional space and shelving units the warehouse would need in 2025 to be at various levels of space utilization, specifically looking at the additional amount of pick area space needed in a new warehouse. We recommend moving into the warehouse at 75% full capacity, which will require a 26,000 sq. ft facility with a 15,200 sq. ft for the picking area .
We also performed an analysis to find the optimal location in Atlanta to obtain a new warehouse. We inputted all the shipping routes they run every week and used Excel solver to calculate the optimal location for a new warehouse. This optimal location is about 8 miles northwest of DIW and is very close to the new hospital location.
Deliverables
For our future demand forecast, we will provide our 8 forecasting formulas derived from the regression equations. Next, we’ll provide a diagram with our item reorganization strategies as well as an outline of how to optimally place items in the new warehouse based on our Bin Optimization Model. We will also provide a list of updated par values along with the methodology used so CHOA can re-calculate par values in the future. This deliverable goes along with an Excel Macro that identifies stagnant and overstocked items in the warehouse along with a set of usage instructions
As far as implementation, our growth predictions from our demand forecast models will be used by the client to make decisions on when and how to scale up operations. For reorganization, we recommend the client use our provided methodology to determine how to optimally place items in their new warehouse. For par values, we recommend that CHOA begin using our recalculated par values for reordering and then use our provided calculations to recalculate par values every few years to ensure they reflect current demand. Finally, we recommend that CHOA use our macro annually to evaluate current stagnant and overstocked items and whether they should be removed from the warehouse.
Project Information
Student Team
Carolyn Callaham, Susannah Cole, Cristian Corrales, Mallory Greene, Gates Ledbetter, Katherine Markey, Abigail Russo, Emily Teague
Faculty Advisor
Faculty Evaluator
