Contact Twinn

CONTACTER Twinn

  • By Lanner
  • In Blog
  • Posted 31/10/2019

For the food and drink supply chain, the nirvana of all equations is of increasing profit, while responding to demand, and improving efficiencies – and doing this in a way that presents zero risk to the business. Historically, static modeling has helped in certain areas, but with the increased adoption of predictive digital twins, companies can more easily model all stages of the manufacturing process, creating business value and unprecedented insight. From Mars to Britvic, companies across the sector have begun solving their own equations leading to some ground-breaking results.

The status quo

With increased consumer demand, strict safety and nutritional guidelines, the drive for efficiency and corporate reputation, food & beverage companies are under increased pressure to innovate their manufacturing processes. Key to this is modeling – ‘how to do things better’ while increasing (and not risking) efficiency, production and the bottom line (as just three of the major performance measures).

Current methods to this are principally human-based estimations and spreadsheet calculations.

You can build a model in Excel and develop scenarios based on changes in demand estimation and predicted production inputs, but the flaws with non-stochastic calculations (or those that use ‘fixed’ markers) make it a high-risk enterprise. This type of planning solution often suffers from human error, lacks the flexibility to adapt to changing scenarios and hidden underlying logic understood by only a few key experts.

Considering the number of key variables that exist when considering, for example, a confectionary production process across a single site are tremendous – often framing the calculation is far beyond the capability of most individuals and excel modelers. As businesses become more complex, so does the production equation. Simplification to find a solution (often through treating key variables as static averages) can also have critical implications for manufacturing processes and risk management.

As companies become more digitally complex and connected within Industry 4.0, these implications can become game-changers, as Paul Myler, Director of Supply Chain Strategy & Industrial Engineering, Mars Chocolate North America, explained “Static modeling works well in certain areas, but we needed a tool which could explore high volumes of what-if scenarios and plot multiple answers / ranges to achieve an in-depth, contextualised understanding of our six sites, based on variable capacity, demand and product mix.”

Companies can try and test new scenarios in real life but the risks of this in terms of cost, time and repeatability are profound. Imagine experimenting with 'what if?' options on a real production line, or with a new distribution process. KFC anyone? At best it would be impractical and at worse financially destructive and, in certain scenarios, possibly dangerous.

This type of real-life trialing is very much based on the ‘gut feeling’ of professionals. The palette of the individual, who has worked in this sector for decades, can and will understand the changes and variables, and drive change. This potentially can work if you are replacing a spigot in a pipe, but what if you are producing four million bottles of beer across three continents, each with their own climate and individual characteristics? We are inherently biased as a species and that makes us entirely fallible when it comes to predictions. Should you trust your entire business on gut instinct?

Enlightened alternatives

Industry 4.0 has seen businesses driven to embrace digital models that place connectivity, data analytics and stronger customer focus at its heart, and nowhere clearer can this embrace be seen than with predictive simulation.

Predictive simulation enables companies to create digital twins of their processes (from business admin functions through to production and supply chains). These predictive digital twins reflect the specific manufacturing processes of your business, demystifying much of the analytical process through rich interactive visualizations and powerful future state data that unlocks unprecedented insights and foresight of your business performance.

Predictive digital twins incorporate the variability and complexity of real life, taking all important logic and timing information into account for facility processes. In addition to simulation, predictive digital twins can be further augmented with other analytical technologies such as machine learning (ML), Artificial Intelligence (AI) algorithms and dashboard technology (Power BI, Tableau, etc) to enhance data preparation, results analysis and performance optimization.

Virtually risk-free decisions

Being able to test and make assumptions and decisions in a virtual world provides clarity across areas such as capital investments, resource planning, process design or even service policies.

“We were able to home in on what the true questions were that we were trying to answer and that was really valuable to us,” said Neil Brinkman, Operations Optimization Manager, Britvic.

Britvic used Lanner's predictive digital twin to help design and plan a new high-speed manufacturing line at its Leeds facility. Being able to bring their processes to life, the company could confirm running costs and build these factors into the budget which provided invaluable financial data but also buy-in and confidence with stakeholders.

Once a predictive simulation model of a manufacturing or business process is created, you can ask ‘what-if?’ questions of the model and run differing scenarios. This allows both analysts and decision-makers to understand their data, future impact and consequence of each scenario, without incurring any risk or cost.

Case study Mars: Ethical Principles in a Commercial Corporate World

A fundamental overriding challenge for  Mars was driving efficiency throughout the business while maintaining a core corporate principle of manufacturing quality chocolate products in the local markets where they get consumed.

Mars have almost three dozen product lines running across six sites in the USA, with complex internal supply chain pressures with each site with varied chocolate making capabilities, multiple chocolate types and varying product mix and consumer demands.

As part of the project with Lanner, Mars identified that the scale, complexities and interdependencies involved would require modeling capabilities more sophisticated than its existing spreadsheet-based capacity planning systems.

Mars worked closely with Lanner to create a simulation model that predicted supply chain performance based on the above variables. A key focus was understanding the impact of key strategic trade-offs, for example, whether to make all chocolate types at the location it is required or build fewer larger chocolate making facilities and ship products across the country. The simulation model could quantify how best to work within such parameters.

The use of predictive simulation has provided Mars with insight into its existing and planned future operations, identifying current risks throughout the supply chain and highlighting opportunities for cost savings and performance improvements. In turn, this has provided foresight to build business cases for new facilities and justify investments in chocolate making capacity, thus ensuring the right investments are made at the right time.

Food for thought

As we know, the food & beverage industry is faced-paced and ever-changing, with new pressures to adapt and managing change is critical to success. Predictive simulation enables decisions to be reached quickly, without impacting production or supply chain delivery. It provides you with answers to the most fundamental or innovative transformation questions, allowing you to make production decisions that will have a profound impact on your business. Ultimately, simulation leads to smarter risk-free business decisions and drives higher returns on your investments.


Loading blog comments..

Post a Comment

Thank you, your comment is awating approval
Submit