Optimized mixing processes in the food industry

Optimal batch selection for more quality and health

© chitsanupong - Adobe Stock

Enjoyment and quality are essential characteristics for any food. The variety of flavours is achieved in production by mixing batches with a wide range of ingredients. On the manufacturer's side, the batches in stock must be prepared in advance for the mixing process as part of production planning. This is the only way to ensure the quality of the mixtures for the customer orders on the one hand and to use the stock sustainably on the other.

In this context, Fraunhofer IIS is using mathematical optimisation methods to develop a batch selection in homogeneous mixing processes that enables more efficient and sustainable stock management.

Optimised production planning for efficient inventory control

The versatile planning of blending processes are core components of food and beverage production. The main challenge in production is to use the right batches for blending, depending on the origin, cultivation of the food ingredients and the method of further processing. Quality parameters, such as value-giving ingredients, are strictly measured and tested in a laboratory analysis and have an influence on the final mix, because a legally prescribed framework has to be adhered to.

The batch inventory determines which orders can be formulated. Currently, batches are prepared with a lot of manual planning effort under a wide variety of criteria. Consequently, it makes sense to link the decision as to which orders are to be formulated with the batch selection. The aim of the project is therefore to develop a decision-supporting software planning tool that considers the following central questions in an integrated way.

How can batches be selected during formulation in such a way that

  • quality standards and requirements are ensured, taking into account the measured values from laboratory analysis during and after mixing?
  • sustainable and forward-looking stock planning is made possible?
  • customer orders can be served in the best possible way with the available stock?

Planning software has been developed for the final planning step from semi-finished product to finished mix. The basis is a customised mixed-integer optimisation model. This model was implemented in a software whose test runs promised high added value for production planning.
With comparison of planning strategies and flexible interfaces to a user-friendly software solution

In this project, the underlying mathematical model was formalised with additional production rules and planning practices. In the process, information on customer orders, stock, laboratory analysis for development and testing was made available by the project partner. This makes it possible to integrate the existing software into the project partner's IT infrastructure via flexible interfaces in order to seamlessly integrate decision support into everyday operational planning.

Potentials for even more resource efficiency

The software tool, which is based on mathematical optimisation, flexibly generates a batch proposal for the mixture on the basis of existing requests for products. An expansion option to further increase the resource efficiency of warehouse planning is to extend the planning from one production step to the entire production chain from raw materials to semi-finished goods to the finished mixture. Further future potential can also be found in the integration of demand forecasts to estimate orders that have not yet been placed, which could be seamlessly fed into the existing tool.

You might also be interested in

 

OBER – Optimized warehouse planning

Inventory management is a key challenge for SMEs in the wholesale sector. Although forecasting models are already used there in some cases to predict sales, they are often not state of the art. The Fraunhofer Supply Chain Services Group at Fraunhofer IIS is therefore developing and implementing a process to quantify the uncertainty of a forecast using state-of-the-art AI methods and, based on this, automatically determine optimal inventory levels.

Coffee as a driver of innovation

At Fraunhofer IIS, coffee becomes science alongside tea. Read more about the series »Innovation Driver Coffee« in the Fraunhofer IIS magazine here.

 

Field of research

AI-based demand forecasts for logistics, trade and transport

We bring AI-based demand forecasting to bear in logistics, retail, and manufacturing to improve predictions and quantify forecast uncertainties.

 

ADA Lovelace Center

Self-optimizing adaptive logistic networks

Tea recipe planning is a multi-step process in a logistics ecosystem.At the ADA Lovelace Center, diverse AI methods are being investigated in the »Self-Optimizing Adaptive Logistic Networks« application.

 

ADA Lovelace Center

Mathematical optimization

Mathematical optimization is an indispensable component of model-based decision support by providing planning solutions in a variety of domains. In the competence pillar of the ADA Lovelace Center, extensive existing expertise is further developed in several promising directions, namely real-time planning and control.

 

White Paper

Data Analytics für die Supply Chain

Get an overview of common analytics methods and their use cases in supply chain management here (german).