There for The Picking
ISE Magazine –Volume 48: Number 08
By Erkut Sonmez, Deishin Lee, Miguel I. Gómez, Xiaoli Fan, Laurie Caldwell, Natasha R. Thompson and Melissa Knowles

Supply chain management is the practice of getting the right thing to the right place at the right time. And in a world that increasingly pays attention to sustainability, what could be more important to get to the right place at the right time than food?

Meanwhile, as much as one-third of total agricultural production in the United States goes uneaten every year; that’s counting the losses all along the supply chain, from farm to plate. Focusing just on point-of-origin (farm) loss, one study estimated that 6 percent to 7 percent of the crops planted in the United States go unharvested each year. Reasons for this include deflated market prices at the time of harvest, rejection by wholesale buyers due to slight imperfections (apples, say, that have a brown spot but are still perfectly edible) or the fact that harvesting machines just don’t catch everything.

Many hunger relief agencies have started to address the dual problems of food insecurity and food waste despite the inherent challenges they face in managing produce supply chains, such as delivering the goods before they spoil and maximizing the nutritional content and value.

The biblical practice of gleaning seems a natural solution to this dilemma of too much here but not enough there. The idea was that a wealthy farmer would allow the poor to come onto his land to gather up whatever had been left behind after harvesting.

In a partnership between a food bank, gleaners and academics, we have applied analytical tools to improve gleaner volunteer management.

Gleaning: Waste not, want not

In setting up a successful gleaning organization, there are a number of things one must take into account with respect to the local community and its food shed: farm location, median farm size, crop types, farmer outreach and education, liability, transportation infrastructure and formal collaboration with distribution partners such as food pantries, but also face the same challenges that farmers do in maximizing labor efficiency, but with the added challenge of working with volunteers. When a farmer offers a chance to glean, the right number of volunteers must be organized at the right time and at the right place.

In other words, it’s an operations problem, one suited to traditional industrial and systems engineering tools. But this time, the logistics problem is scheduling volunteer labor– getting the people to the products rather than the other way around.

The Uncertainties of Gleaning

We formed a partnership between the Food Bank of the Southern Tier (FBST), the Boston Area Gleaners and academics interested in operations management, agricultural economics and sustainable food systems to see what we could do to improve gleaner volunteer management, using the analytical tools we apply to operations management. In particular, we developed a stochastic optimization model that represents volunteer gleaning operations. Stochastic optimization is a mathematical technique for revealing the best decision even when there is uncertainty in the factors that will determine the outcome. Gleaning organizations are ideal candidates for this technique: They face unpredictable factors as they try to maximize the amount of produce they can put in the hands of the needy.

Two critical sources of uncertainty are:

  • Uncertainty in donations.
  • Uncertainty in the workforce.

What to Model

Our objective was to determine the volunteer gleaning schedule that would maximize the amount of food gleaned, using the staff, volunteers and trucks the Food Bank of the Southern Tier had available.

A typical gleaning trip lasts about three hours. The food bank announces a gleaning foray and volunteers decide whether to show up at the scheduled place and time. Key decisions for the Food Bank of the Southern Tier are how many gleaning trips to schedule and when. A regular schedule would make it easier to schedule resources, such as staff and trucks, so our optimization focused on choosing the best regular schedule for gleaning trips.

In principle, a schedule might consist of anywhere from one to seven days a week. If it were, say, three days, those days might be Monday, Tuesday and Saturday, or they might be Monday, Wednesday and Saturday or some other combination.

Using U.S. Census data combined with information from the Boston Area Gleaners on their gleaning processes, we characterized the arrival rate of gleaning opportunities (donations) and the distribution of volunteer gleaner attendance. Understanding process constraints such as gleaner productivity and gleaning opportunity windows – so many days to get out there and glean before it was too late – was critical to formulating the model.

What We Learn

We used a simulation-optimization approach to solve the model. We ran thousands of simulations testing different weekly schedules and performed sensitivity analyses on various farm and volunteer gleaner characteristics.

As expected, some patterns appeared, and we could draw what we felt were practical conclusions that would help the Food Bank of the Southern Tier to gather as many apples as possible given its limited farms, trucks, time and volunteers.

While our findings are specific to the Food Bank of the Southern Tier at a particular time, the qualitative insights can be applicable to other gleaning operations – and, in fact, to any organization that depends on volunteers.

For example, does increasing the number of gleaning trips per week increase the number of apples gleaned? What reason can there be to turn down free food? What about the number of participating farms?

In short, the simulation optimization approach can be used to test different scenarios in complex stochastic settings. Gleaning operations is one such setting that can benefit from this approach. Although it is discouraging that there are still so many food-insecure people in the United States, it is encouraging that there is so much good food potentially available for the picking. To make the most of limited volunteer resources, techniques such as stochastic optimization modeling also are there for the picking.