ISSUE

Some products require strict packaging to maintain their effectiveness, especially vaccines and other pharmaceutical products. To ensure that the cold chain is not broken (temperature excursions during which the product is exposed to temperatures that can degrade its quality), our client conducts extensive simulations for each customer’s journey.

CUSTOMER

Lain Pharma, Emball’ISO Lyon (FRANCE)

SOLUTION PROVIDED

Neovision has been approached to develop packaging optimization software based on variations in thermal parameters. We have developed a solution capable of quickly predicting temperature variations and potential temperature excursions.

CUSTOMER BENEFIT

This solution allows operators to choose the most suitable packaging based on the climatic conditions the product will encounter during its journey. It helps to minimize the risks associated with the cold chain, optimize packaging selection, and optimize associated routes.

RÉALIZATION

Multiple models have been used to optimize thermal predictions. Initially, our client used physical simulation models to simulate different temperature variations based on the interactions between objects. However, these simulations were time-consuming. To overcome this limitation, Neovision developed learning models based on historical time series data of interior and exterior temperatures of packaging. These models enable the prediction of thermal variations for new journeys.

Data, Anonymization, Annotations

The operator selects a parameterized package model within the web application (size, cooling modules, etc.). Then, they input the estimated itinerary of the package along with the associated temperatures. For example, for a package traveling from Berlin to Grenoble, they could provide the following information:

2 days of storage in Berlin – estimated outside temperature: 5°C
1 day of air transportation from Berlin to Lyon – estimated temperature: -10°C
1 day of transportation from Lyon to Grenoble – estimated temperature: 10°C

The web application generates a graph for each packaging option, displaying the interior temperature of the package during the journey, as well as specific moments when the product will be exposed to unsuitable temperatures. This allows the operator to adjust the packaging models in the interface and find the most suitable packaging option.

TECHNOLOGIES & ASSOCIATED EXPERTISE

PREDICTIVE ANALYSIS, DEEP LEARNING, WEB APPLICATION