Content area
SHIELD Illinois launched a statewide, saliva-based COVID-19 testing network that reaches 1,700 schools, colleges and universities in Illinois. The depot was essential to this operation; it served as the central distribution hub daily routing thousands of samples to testing laboratories across the state. Depot operations were heavily impacted at the peak of their pandemic, owing to an extremely erratic arrival pattern of samples and a manual dispatching system, thus often leading to inefficiencies and longer cycle times. This case study simulates routing strategics, assessing their impact on travel times, resource requirements and cycle times from collection to results. Through simulations tests, this study aims to obtain optimal dispatch strategies that will make operation more efficient and consistent thereby minimizing delays. The findings of this research will provide valuable insights into future logistics for large-scale public health responses to epidemics.
Abstract ID: 6412
Abstract
SHIELD Illinois launched a statewide, saliva-based COVID-19 testing network that reaches 1,700 schools, colleges and universities in Illinois. The depot was essential to this operation; it served as the central distribution hub daily routing thousands of samples to testing laboratories across the state. Depot operations were heavily impacted at the peak of their pandemic, owing to an extremely erratic arrival pattern of samples and a manual dispatching system, thus often leading to inefficiencies and longer cycle times. This case study simulates routing strategics, assessing their impact on travel times, resource requirements and cycle times from collection to results. Through simulations tests, this study aims to obtain optimal dispatch strategies that will make operation more efficient and consistent thereby minimizing delays. The findings of this research will provide valuable insights into future logistics for large-scale public health responses to epidemics.
Keywords
Simulation, depot, dispatch, disease testing, transportation
1. Introduction
During the peak of the COVID-19 outbreak, Illinois needed an expedited large-scale testing solution. Shield Illinois ("SHIELD") was established by the University of Illinois Urbana-Champaign (UIUC) with funding from the Illinois Department of Public Health to administer UIUC's innovative saliva-based test in over 1,700 educational institutions including K-12 schools and universities. SHIELD aims to minimize turnaround times for test results, thereby ensuring timely information for public health decisions. SHIELD'S logistics network consists of all schools serving as sample collection points, laboratories that process the tests, and a distribution center ("depot" located in Darien, IL) where samples are sorted and rerouted to labs. Thousands of saliva samples are processed through the depot every day. Due to the random nature of sample arrivals and a largely manual rerouting process that was reliant on spreadsheets and ad-hoc channels, depot operation was under massive pressure at the peak of the pandemic. This case study aims to improve the depot's operations by constructing a simulation-based model to evaluate various scenarios.
2. Problem Statement
Although SHIELD'S logistics network successfully responded to a high-volume demand, day-to-day depot operations continued to rely on manual and qualitative decision-making processes when it comes to when and how to send samples to different labs, relying on the depot manager's experience, ad-hoc communications with labs, and rules of thumb built up over time. Such approaches produce suboptimal results: samples could waste time at the depot, lab workloads are unbalanced and peaks in testing demand may be underprepared for. This delays the turnaround of test results and compromises the network's ability to provide timely test results.
3. Background
SHIELD'S depot aggregates saliva samples from different sites every day, verifies the manifests and forwards the samples to the labs. Depot operations historically used manual and qualitative decision-making methods, primarily determined by a mix of direct conversations with laboratories; managers' personal insights and established rules of thumb. This often results in delays, workloads at labs that are not balanced, and longer cycle times from sample collection to result reporting. To address these challenges, the dispatching strategies will be examined through a simulation-based modeling approach for analysis and improvement. Simulation provides a controlled virtual environment in which different dispatch rules, resource allocations, and operational changes can be investigated, allowing for rigorous and fair comparisons between competing solutions.
4. Approach
The simulation model (Figure 1) is built in SIMIO to focus on the period from January 14 to May 5,2022, a period that coincided with the advent of the Omicron variant. The Source Arrival Process is responsible for generating incoming entities based on a predefined Arrival Table, ensuring that arrivals follow a structured schedule. Once generated, entities flow into the depot, where they are processed and routed to labs (lab names: SHIELD451, GOH, ISU, LOYOLA, MADISON, and UICT3). The dispatch process is based on the Dispatch Pattern Table (made from SHIELD'S dataset). For instance, the first delivery most frequently headed to Loyola at 13:00 (43% of days), averaging 2,200 items over 75 days. Inventory levels are verified to determine whether there is sufficient stock. If the stock is insufficient, the system logs the shortage by incrementing the NotEnoughlnventory counter. In this study, inventory refers solely to the depot's on-hand supply of saliva samples awaiting dispatch. Shipment quantities are assigned based on real-time stock availability or predefined dispatch rules, such as specific location-based allocations (e.g., LOYOLA at 1:00 PM requires 50-100 units). A key feature of this model is its sophisticated inventory control mechanism, which checks stock at the distribution center whenever a dispatch order is generated. We implement three distinct stock allocation scenarios: Full Quantity Dispatch (sufficient stock to fulfill the entire order), Partial Quantity Dispatch (only part of the request is available) and Complete Stock Clearance, a special scenario where all remaining stock is dispatched completely and efficiently, thus finalizing all inventory operations.
A priority-based decision system is developed to manage these scenarios effectively. The model continuously monitors stock levels, maintaining a shortage counter. Entities follow a dispatch process where quantities adjust based on availability; partial dispatch occurs if demand exceeds stock. Real-time updates to assignment variables prevent imbalances. Before finalizing dispatches, all constraints are verified to ensure compliance with shifting demand. If any requested quantity exceeds available stock, the notEnoughlnventory counter increments and dispatch adjusts to current levels, preventing negative inventory situations. The Depot Routing and Destination Processing Model integrates travel logic, dispatch scheduling, and destination handling to ensure efficient sample movement. The TimePath logic dictates unidirectional travel with a First-In-First-Out (FIFO) ranking and references the TravelTimes table (which is made from the average times taken to travel between depot and each lab) to determine shipment durations dynamically. The decision ensures all logical conditions are met before terminating the process. Each laboratory is represented with a fixed throughput capacity, it can process only a given number of samples per day. The model calculates a per-minute operational cost of $2.10 for trucking by integrating ATRFs cost components and operational time data. The simulation model is validated by a comparison of average waiting time over 75 days. The depot experimented with various dispatching strategies, adjusting from a two-delivery-per-day schedule to a ten-delivery-per-day approach to optimize operations. The seven-delivery-per-day pattern, shown in the provided image, was one of the tested strategies. This schedule included dispatches at 12:00 AM, 12:00 PM, 1:00 PM, 2:00 PM, 3:30 PM, 5:00 PM, 6:10 PM, and 7:10 PM, with Loyola acting as the final recipient for any remaining samples. The observed waiting time during the study period amounted to 111 minutes, which aligned closely with the simulated average waiting time of 111.1 minutes. The experiment showed minimal result differences thus validating the accuracy and trustworthiness of the simulation model.
5. Solution
Scenario 1 follows a six-delivery-per-day pattern, developed through historical analysis by adjusting time and quantity while maintaining lab capacity limits. The schedule was optimized to balance workloads, reduce waiting times, and ensure steady processing efficiency. This structured approach helps prevent lab overloads while maintaining consistent dispatch operations.
Scenario 2 follows an eight-delivery-per-day pattern, prioritizing closer labs to reduce costs while maintaining lab capacity limits. Samples are sent earlier, starting at 12:00 PM, to improve processing efficiency. This approach balances cost and turnaround time by optimizing dispatch frequency and lab proximity.
Table 1 compares two dispatching strategies.
Scenario 3 uses weekday-dependent scheduling for 75 days by executing predefined delivery sequences. The depot follows precise delivery patterns on each weekday that they repeat weekly. The dispatch sequence for each weekday is as follows:
Monday: LOYOLA GOH SHIELD451 ISU MADISON UICT3
Tuesday: SHIELD451 ISU MADISON UICT3 LOYOLA GOH
Wednesday: ISU MADISON UICT3 LOYOLA GOH SHIELD451
Thursday: MADISON UICT3 LOYOLA GOH SHIELD451 ISU
Friday: UICT3 LOYOLA GOH SHIELD451 ISU MADISON
This structured approach allows the system to evenly distribute resources throughout the week while optimizing dispatch efficiency. However, the average waiting time in this scenario is 73 minutes.
To determine the most suitable scenario for the depot, we evaluate dispatch efficiency, average waiting time, and transportation cost for all three scenarios. The depot should select Scenario 1 to achieve the best cost performance. The reduction of waiting time demands Scenario 2 as an efficient solution. The choice of Scenario 3 proves suitable when operational stability alongside structured planning stands as the main organizational priority. Scenario 2 stands out as providing the most optimal solution by reducing waiting times with controlled transport expenses.
6. Conclusions
The research investigates dispatch optimization approaches at SHIELD depot that requires improvement due to its manual scheduling processes and unpredictable sample delivery times. The model used simulation to analyze three dispatch strategies for evaluating waiting time performance as well as transportation expenses.
Acknowledgements
This study is sponsored by SHIELD Illinois (Awards 109753, 113832; NIU Grants G3A63106, G3A63118).
Copyright Institute of Industrial and Systems Engineers (IISE) 2025