Engineering that Matters: A Summer Immersion at The Foodbank

Manav Kandhari
Department of Engineering Management, Systems and Technology
The University of Dayton
Dr. Kellie Schneider
Department of Engineering Management, Systems and Technology
The University of Dayton
Dr. Corinne H. Mowrey
Department of Engineering Management, Systems and Technology
The University of Dayton

Households that do not have consistent access to a sufficient quantity of affordable, nutritious, and culturally appropriate food are said to be food insecure. In 2018, the USDA reported that approximately 11 in 99 Americans were facing food insecurity, or more than 37 million Americans, including more than 1111 million children (age 18 and younger) [1]. Feeding America, the nation’s largest domestic hunger-relief organization, tracks food insecurity nationwide. In a 2018 analysis, Ohio ranked 1111th in the nation with a food insecurity rate of 13.9% [2]. With the devastating tornadoes ripping through the area on Memorial Day in 2019 and the ongoing COVID-19 pandemic, Feeding America estimates that Ohio’s food insecurity rate for 2020 has increased to 18.1% [3].

In an attempt to combat food insecurity, the U.S. government has implemented many different programs over the past eight decades. However, many households still rely on the private food assistance network to meet at least some of their nutritional needs [4]. Within the private food network, food banks serve as central warehouses that collect, purchase, store and distribute food. The Foodbank. Inc. serves as the primary source of food for the Dayton, OH area network. Annually, The Foodbank distributes more than 17 million pounds of food through a network of more than 100 partner agencies, and were recently recognized as one of America’s best food banks. The Foodbank’s vision that “no one should go hungry” drives them to constantly strive for continuous improvement. Through an ongoing partnership with the University of Dayton, The Foodbank has focused on using community-based operations research (CBOR) to improve the efficiency of their truck routes.

Community-based operations research is defined as a set of analytical methodologies used to solve problems where the interests of underrepresented, underserved, or vulnerable groups in the local area are prioritized [5]. CBOR makes use of statistical models and analytical methods to address problems of core concern to the individuals, government, and non-governmental organizations seeking to address and work towards the issues of social inequity. Community-based problems place significant emphasis on the needs of specific populations in well-defined neighborhoods. The local nature of this work exerts a strong influence on the analytic models and resulting policy or operational decisions. This work is often considered “messy” as the formulation and solution of these problems are highly dependent on political and social considerations [5]. Examples of CBOR can be found in human services (humanitarian logistics, libraries and literacy, education and family support services), community development (housing, community/urban planning, transportation), public health and safety (health care, criminal justice, emergency services, hazardous/undesirable facility location, food insecurity) and non-profit management.

One of the most common problems encountered in transportation logistics is the vehicle routing problem (VRP). The VRP aims to minimize the total cost of transportation by finding optimal routes for multiple vehicles visiting a set of locations. The VRP generalizes the well-known traveling salesman problem (TSP) and was first introduced by George Dantzig and John Ramser more than 50 years ago [6]. Some examples of VRPs arising in urban transportation are the efficient routing of school buses [7], and the design of tourist tours to visit multiple points of interest in a city [8].

Although there are many software packages available to solve the vehicle routing problem, we specifically chose an open-source solver for our application. The VRP spreadsheet solver, introduced by Güneş Erdoğan, was specifically chosen because of its interface, ease of use, accessibility, and flexibility to run on any operating system with Microsoft Excel [9]. The VRP spreadsheet solver has been used by organizations in diverse sectors, but its application is most straightforward in the field of logistics. Our desire to use this solver grew stronger as a result of its tremendous potential to achieve savings by increasing efficiency, as well as its ease of handling computational complexity for small to medium-sized enterprises, like The Foodbank. The scope of our application of VRP in CBOR is limited to the U.S., as characteristics, including laws and regulations, may differ from country to country around highly sensitive issues such as poverty, food security, and homelessness.

Improving Retail Routes
The Foodbank’s mission is to relieve hunger by acquiring and distributing food through a network of partner agencies. After a series of tornadoes devastated the area on Memorial Day weekend of 2019 [10], followed almost immediately by the global pandemic, The Foodbank faced operational hurdles as they worked to manage these crises. As a result, The Foodbank was forced to respond extemporarily to the increased need for food throughout 2020, focusing more on the effectiveness of their work rather than efficiency. The pandemic response made The Foodbank’s work particularly challenging, as non-essential personnel was asked to stay home and volunteer services were suspended. The Foodbank’s operations, however, achieved some stability as they received extra help from the National Guard who assumed many of the day-to-day operations [11]. With the widespread availability and distribution of COVID-19 vaccines, The Foodbank began to return to pre-pandemic operations. Personnel and volunteers were allowed to return to the warehouse and The Foodbank was able to return their focus on improving their operations. Because the University of Dayton (UD) has a long-standing history of working with members of the Dayton community, The Foodbank reached out to UD’s School of Engineering and ETHOS Center for help. The Engineers in Technical Humanitarian Opportunities of Service-Learning (ETHOS) program is open to all engineering students at the University of Dayton. ETHOS seeks to provide service-learning experiences for students so that they may gain perspectives on how engineering and technology are influenced by varying conditions around the world. Participating students serve others by applying their engineering knowledge and skills for humanitarian purposes [12]. Working collaboratively, the ETHOS Center and the Department of Engineering Management Systems and Technology (EMST) sought to improve one part of The Foodbank’s operations: its retail collection routes.

One of the primary ways The Foodbank acquires food is through donations from local retailers. Using a fleet of privately owned trucks, The Foodbank visits more than 4040 retailers on a weekly basis to acquire donated food. The impacts of the Dayton tornadoes and the ongoing response to COVID-1919 resulted in unbalanced collection routes that were formed ad hoc. The overall goal of this project was to increase the efficiency of collection routes by reducing the cost and time required to collect retail donations. The uniqueness of this project is highlighted by the fact that our work is community-focused and includes a component that specifically addresses an equitable distribution of work among all trucker drivers. The project’s geographic scope was limited to the area served by The Foodbank, namely the region of Pebble, Greene, and Montgomery County. Our analysis of the data provided by The Foodbank allowed us to identify that multiple stakeholders need to be considered when evaluating the trade-offs between efficiency, effectiveness, and equity in our solutions. The primary stakeholders shortlisted are Donors (partner agencies, govt., non-governmental organizations, individual donors) and Service providers (employees, partner distributors, drivers, volunteers, etc.), whereas the list of secondary stakeholders includes Partner agencies and Customers (individuals/families receiving food from different programs).
The team utilized the following three steps to conduct the project:

  1. Understanding the project activities
    This included reviewing existing retail routes to establish baseline parameters such as the number of routes required for the collection, the number of collections per week, and the total number of miles driven by each truck. The resources for our project were limited to a fixed number of trucks and drivers, as well as numerous time constraints that had to meet the retailer’s expectations. Our project was responsible for reducing both the cost and time spent on truck routes. Upon further investigation, we discovered that the cost savings were a direct result of reducing the number of miles driven to cover the same number of stores by the truck, but the time to cover those stores was not, resulting in increased computational complexity. Travel time is considered highly variable since it is subject to a variety of unknown factors such as traffic, availability of drivers, weather conditions, etc.
  2. Establishing and evaluating baseline
    The baseline evaluation of existing retail routes is based on the route sheets provided by The Foodbank at the start of the project in May 2021. Close collaboration with operations workers allowed for the integration of new retail donors into collection schedules, as well as recognition of each donor’s preferences for collection days and hours. At baseline, there were 47 retail donors, and 44 trucks were used to make 126 collections per week, as summarized below in Tables [table1] and [table2]. The research team worked closely with operations managers and truck drivers to understand the various restrictions and preferences of retail donors. For the 19 existing routes, a total of 926.08 miles were driven each week.

    Table 1: Number of retailers visited per week by day of the week and by truck number.

    Truck Monday Tuesday Wednesday Thursday Friday
    1 7 4 8 6 9
    2 9 7 7 5 9
    3 8 6 8 0 9
    4 8 1 2 5 8
    Total 32 18 25 16 35

    Table 2: Distance traveled in miles per week by day and by truck number.

    Truck Monday Tuesday Wednesday Thursday Friday
    1 35.7 64.5 41.63 57.71 59.92
    2 46.76 41.89 46.90 28.52 46.76
    3 42.59 48.44 57.44 n/a 44.69
    4 53.15 24.88 36.54 65.35 82.7
    Total 178.2 179.71 182.51 151.57 234.07
  3. Optimization
    The VRP spreadsheet solver was used to optimize retail routes on a daily basis [9]. Various settings in the VRP spreadsheet solver can be adjusted – such as the shortest distance vs fastest time, the truck capacity, the work start time, and the number of trucks – in order to create a series of proposals for the operations team to consider. A snapshot of the input to the VRP solver for the accepted (proposal) route on Tuesday is shown below in Figure [Figure 1].


    Figure 1: The input sheet for the VRP solver.

The next step in the process was to input the stores to be picked up and their corresponding street addresses. The VRP solver identified the coordinates (latitude and longitude) of each location using Bing maps. These coordinates were used to calculate the distance and travel time, as well as to facilitate the visualization of the solution. Once the coordinates were located, we input the different time constraints associated with each store in order to satisfy the preferences of the various retail donors. An approximate service time for each retailer was determined using driver feedback as well as data from previous projects. Figure [Figure 2] shows an example of a Tuesday location sheet with accurate pickup addresses, coordinates, and time constraints corresponding to each store. A map visualizing all Tuesday pickup points using the input coordinates is shown in Figure [Figure 3].


Figure 2: The location sheet for the VRP solver.


Figure 3: The map of the pick-up points for a Tuesday.

During the route optimization, we used different combinations of different input constraints, including vehicle capacity, number of vehicles, driving and working limits, etc. to obtain and identify several feasible solutions. The research team then discussed the results with The Foodbank operations team to select the best set of routes. The vehicle input sheet for the accepted proposal is shown below in Figure [Figure 4].


Figure 4: The vehicle input sheet for the VRP solver for a Tuesday.

The key features of the VRP solver include its ability to calculate the distance between each pickup point using Bing maps and its ability to measure the travel time between each pickup point based on the specified average vehicle speed. Figure [figure5] shows a snapshot of Tuesday’s distance sheet that includes the distance and duration between all listed stores.


Figure 5: The distance sheet for the VRP solver for a Tuesday.

After a lengthy selection process, our team was able to finalize the optimized proposals for the Tuesday, Wednesday, and Thursday routes, which not only met the given constraints but also matched the expectations of all drivers and supervisors. The key information on the optimized routes is provided below. A visualization of the baseline routes compared to the optimized routes for each day is shown in Figures [Figure 6]-[Figure 8].


Figure 6: Tuesday’s current route map (left) and its proposed optimization (right). Total miles reduced = 24.98, Total route time reduced = 60 minutes.


Figure 7: Wednesday’s current route map (left) and its proposed optimization (right). Total miles reduced = 27.09, Total route time reduced = 20 minutes. The expectation of creating a training route (green lines) was also successfully fulfilled.


Figure 8: Thursday’s current route map (left) and its proposed optimization (right). Total miles reduced = 26.91, Total route time reduced = 45 minutes.

Future Work
Due to the dynamic nature of fleet operations, it is necessary to occasionally update retail routes. The research team is pleased to continue collaborating with The Foodbank on improving operations. There may be opportunities in the future to further improve the efficiency of their fleet by optimizing delivery operations and other types of donation pickups (such as food barrels) that were not considered in this project. Another significant way to improve operations at The Foodbank is to train members of the operations team to use the VRP spreadsheet solver, giving them the ability to update routes on-demand, in real-time, in order to maintain their commitment to continuous improvement.


Figure 9: The Foodbank truck at one of the retail stores, ready for pickup.

Manav’s Immersion Reflection
In both my academic and professional life, I have been consistently praised as focused by my professors and peers. Whether working on academic, extracurricular, or professional projects, I apply proven communication, management, and problem-solving skills, which I aimed to continue over the summer at The Foodbank. I think of myself as a combination of an imaginative and practical learner, which has served me well at The Foodbank. By conducting research, communicating to obtain feedback from my team, improving my optimization skills, and finally working with several solutions to find one that can meet both the standards and preferences of the operating personnel, I was able to combine real-world experience with theory while gaining hands-on experience. My approach to personal development during this immersion was based on the “3C’s of the entrepreneurial mindset model” [13], namely Curiosity to learn and understand the work culture of a nonprofit organization, developing Connections to expand my professional network, and Creating value by applying my technical skills to support the overall growth of the organization. The photo of our Foodbank team [Figure 10] was taken at the ETHOS Poster Presentation Day, where I was able to offer an example of how engineers can have a significant impact on society.


Figure 10: Our team at the ETHOS poster presentation at the University of Dayton.



  1. A. Coleman-Jensen, M. P. Rabbitt, C. Gregory, and A. Singh, “Household Food Security in the United States in 2018. Economic Research Report No. 270.,” United States Dep. Agric. Econ., no. September 2019.
  2. Feeding America, “The Impact of the Coronavirus on Food Security,” Feed. Am., vol. 140, no. 2, pp. 1–3, 2020, [Online]. Available: Link 
  3. N. Smith, “Ohio projected to experience 18.1% food insecurity for 2020,” 2020, [Online]. Available: Link 
  4. K. Schneider and S. G. Nurre, “A multi-criteria vehicle routing approach to improve the compliance audit schedule for food banks,” Omega (United Kingdom), vol. 84, pp. 127–140, 2019, doi: 10.1016/
  5. M. P. Johnson and K. Smilowitz, “Community-Based Operations Research,” OR Tools Appl. Glimpses Futur. Technol., no. June, pp. 102–123, 2007, doi: 10.1287/educ.1073.0035.
  6. G. B. Dantzig and J. H. Ramser, “The Truck Dispatching Problem,” Manage. Sci., vol. 6, no. 1, pp. 80–91, 1959, doi: 10.1287/mnsc.6.1.80.
  7. T. Bektaş and S. Elmastaş, “Solving school bus routing problems through integer programming,” J. Oper. Res. Soc., vol. 58, no. 12, pp. 1599–1604, Dec. 2007, doi: 10.1057/palgrave.jors.2602305.
  8. D. Gavalas, C. Konstantopoulos, K. Mastakas, and G. Pantziou, “A survey on algorithmic approaches for solving tourist trip design problems,” J. Heuristics, vol. 20, no. 3, pp. 291–328, 2014, doi: 10.1007/s10732-014-9242-5.
  9. G. Erdoğan, “An open-source Spreadsheet Solver for Vehicle Routing Problems,” Comput. Oper. Res., vol. 84, pp. 62–72, 2017, doi:10.1016 j.cor.2017.02.022.
  10. H. News, L. Live, V. Romo, and B. Chappell, “’ Tornado Outbreak’ Devastates Ohio Communities with Winds Up To 140 MPH,” 2019, [Online]. Available: Link.
  11. D. News, Davero Bogart, “Ohio National Guard unit helps food distribution at The FoodBank,” 2020, [Online]. Available: Link.
  12. B. R. Source, M. E. Journal, M. Education, and S. Url, “Re-conceiving the e t,” vol. 91, no. 1, pp. 33–37, 2010, [Online]. Available: Link.
  13. A. G. For and E. Mindset, “The Framework A Guide For Entrepreneurial Mindset.”