用户案例集锦

Forecasting Fruit Freshness with Simulation Apps

Swiss research organization Empa built models and compiled a simulation application that feeds results into a smartphone app used by farmers and traders to predict the shelf life of fresh fruits and vegetables. The work is in support of a multinational coalition whose goal is to improve the use of refrigerated food storage in developing countries.

By Rachel Keatley
September 2023

The post-harvest journey of fresh produce is a notably weak link in the global food supply chain. Each year, approximately one-third of the food produced for human consumption worldwide is lost or wasted (Ref. 1). Refrigerated distribution and storage problems play a major role in these losses, especially in developing countries like India, where a mere 6% of food production enters the refrigerated "cold chain" (Ref. 2) — leaving it vulnerable to decay. Currently, the scarce refrigerated space, such as small, solar-powered cold storage rooms (Figure 1), may be occupied by crops that are already past their peak, even as other shipments decay while waiting for access to a cold room.

As part of a multinational consortium of food supply chain stakeholders, the Swiss Federal Laboratories for Materials Science and Technology (Empa) and the Basel Agency for Sustainable Energy (BASE) developed Coldtivate to help alleviate these problems. Coldtivate is a mobile app that informs its users of the cooling and decay process of different types of fruit in real time. It was rolled out to cold storage operators in three regions of India in late 2022 and in Nigeria in 2023, and it is powered by multiphysics simulation. The farmers who use the app do not actually see the underlying multiphysics models or interact directly with the simulation app: they get the results delivered in a way that is actionable to them through the Coldtivate mobile app.

Simulation-Based Predictions to Optimize Food Storage

The Coldtivate mobile app provides data-driven forecasts on the freshness of produce in a cold room. These forecasts are relayed from a simulation app (Figure 2), which Empa built with the COMSOL Multiphysics^® software and its built-in Application Builder.

Simulation apps are easy-to-use interfaces that are made from existing COMSOL Multiphysics models, where the app designer decides which inputs and outputs to display. Simulation apps are ready-to-use as is, but Empa and BASE wanted to relay their simulation app's data in a way that would be most familiar to their end users, primarily farmers. Since mobile apps are widely used in modern agricultural practices (Ref. 3), they developed Coldtivate to act as the user interface for their end users.

"Our COMSOL app runs on the same headless server that hosts the mobile app," explained Joaquin Gajardo, data scientist at Empa and technical colead for the Coldtivate project. This is what enables both the simulation app and mobile app to relay information back and forth between one another. “We have used the Application Builder to automate the updating of simulations based on changing input parameters [in the mobile app],” said Gajardo. This combines the capabilities of Empa's multiphysics model with the convenience of a purpose-built mobile app. For their simulation app to be used in such a way, they first compiled it into a standalone executable using COMSOL Compiler™.

"Without the Application Builder, it would have been impossible to roll out the digital twins into a mobile app and democratize these multiphysics simulations, and their results, to a broader audience of beneficiaries, like smallholder farmers," said Thijs Defraeye, senior scientist at Empa and professor at Wageningen University & Research.

The Model Behind the Simulation App

Ambient temperatures directly influence the shelf life of fresh produce. But while thermometers are found inside most refrigerated storage units, the data they provide is insufficient for predicting how long fruits and vegetables will last.

"We cannot establish the expected lifespan of produce with a single temperature or humidity curve from a cold room," said Defraeye. "The COMSOL model of each crate receives actual temperature and humidity data from the cold storage rooms, enabling frequent recalculation of remaining shelf life." This model is what Empa's simulation app is based on.

In addition, a sensor in one section of the storage space does not necessarily reflect the temperature on the surface of an apple buried inside a crate, especially if that crate was only recently brought indoors.

To provide a fuller picture of how each shipment of produce may be affected by varying storage conditions, Empa is currently expanding the models to model entire shipments of various fruits and vegetables using COMSOL Multiphysics^® (Figure 3). "For this, we need to use a porous media modeling approach to generate actionable metrics from sensor readings," said Defraeye.

The App in Action

"Let's say a new shipment arrives at a cold room," Gajardo proposed. "The operator opens the Coldtivate app on their phone and enters the type of produce, current temperature, days since harvest, and other relevant values. The mobile app then generates a text file containing this information. The values in this text file serve as input arguments for the COMSOL app, which then calculates the expected shelf life. This calculation is then written into an output file, and the remaining quality and number of days are subsequently displayed in the mobile app’s user interface."

Every six hours, the values shown on the Coldtivate app are recalculated with updated forecasts based on the latest cold room temperature data (Figure 4). Farmers can directly check the remaining shelf life of their crates in the mobile app. If they do not have a smartphone, "warehouse operators can notify them of how long their produce will remain fresh in the cold room," said Gajardo. Ultimately, this information will help farmers avoid distress selling and unnecessary disposal of unsold produce.

Team Effort Leads to Trustworthy Data

Defraeye and his team at Empa have devoted years to building their data-gathering and modeling process. "In early 2021, we were already making models [for analyzing produce freshness], but they were not yet rolled out in actual supply chains," Defraeye said. Inquiries from nonprofit global development organizations sparked the effort that led to the creation of the Coldtivate mobile app (Figure 5).

"We were contacted by BASE, which develops innovative business models to help farmers make better use of available resources," Defraeye said. "The idea was to combine a pay-per-use business model with intelligence that could improve access to cooling. To do that, we needed other partners with close links to the people we wanted to help." These partners included cooling companies and other stakeholders in the Indian states of Bihar, Himachal Pradesh, and Odisha, who joined BASE and Empa in an initiative called Your Virtual Cold Chain Assistant (Your VCCA).

"Trust is the key ingredient," said Defraeye. "The simulations allow us to peek inside what happens to produce over time, and the smartphone app puts that info into the hands of people who can use it, such as cold room managers or farmers. The goal is to add transparency at the point of sale, where it can make a big difference. If the farmers and cold room operators can trust the forecast, that helps them trust each other as well."

Coldtivate in the Field

In August 2022, the Coldtivate app was released to selected cooling companies along with training for cold room managers on how to use the app to track produce in their facilities (Figure 6). To date, the simulation-powered app has been piloted in 17 cold rooms, serving more than 300 farmers, who are reporting a 20% increase in their incomes and a reduction of their post-harvest food losses by 20%. Empa and its partners are now working to expand Coldtivate's impact.

Future iterations of the app will deliver additional relevant info, such as market price forecasts, and even predictions of crop quality based on photos of produce. Now, BASE and Empa are working with organizations in Nigeria and the Philippines to bring Coldtivate's benefits to more of the world (Ref. 4).

The scale of global post-harvest crop losses is daunting, but the coalition behind the Coldtivate project has proven that progress is within reach. "We are seeing how access to simulation-based predictive tools can improve access to cooling," Defraeye said. "Now we need more than incremental baby steps; we need bold action to expand our impact. We do this by putting technology into the hands of more people who can use it to make a difference."

Acknowledgments

The development of Coldtivate was funded partially by the data.org Inclusive Growth and Recovery Challenge grant “Your Virtual Cold Chain Assistant”, supported by The Rockefeller Foundation and the Mastercard Center for Inclusive Growth, as well as by the project “Scaling up Your Virtual Cold Chain Assistant”, commissioned by the German Federal Ministry for Economic Cooperation and Development (BMZ) and being implemented by BASE and Empa on behalf of the German Agency for International Cooperation (GIZ). The project team wishes to thank the main project partners, Koel Fresh Private Ltd., Oorja Development Solutions India Private Ltd., and ColdHubs Ltd., for their contribution in testing and fine-tuning the shelf-life model.