Diageo saves millions of pounds using predictive simulation

The challenge 

Identifying the right opportunities for continuous improvement in packaging operations 

Diageo’s Runcorn and Belfast facilities package its world-famous stout and serve major breweries. In total, more than 150 different products are packaged in several different formats, requiring unique routing along the pack lines.  

Given the range of scheduling complexities and delivery expectations, what were the best ways to drive continuous improvement across its packaging businesses? 

To answer that question, Diageo wanted to understand how predictive simulation could help it enhance agility and seize opportunities in an evolving beverage market. It turned to Twinn and our Witness software, formerly under the Lanner brand, to help achieve this in line with best practices. 

The solution 

A virtual model of packaging facilities that could be customised for the range of equipment, products and production schedules 

Predictive simulation involves creating virtual models of assets and operations so you can experiment with ‘what-if’ scenarios and validate decisions in a risk-free environment. Given the complex, dynamic variables involved in Diageo’s packaging processes, it was the ideal solution for understanding ways to optimise performance.  

Our experts recommended building a generic model that would be readily configurable to represent the different production lines across the packaging business. We began with a study to determine the optimal scope and level of detail – to ensure the model was both easy to use and had the depth required for effective decision-making. We also developed the model so Diageo staff could apply it to all packaging lines with minimum training.  

The Witness model supports a fully flexible equipment network and product mix. Diageo users select the equipment on their site, and it leverages detailed data on speeds, sensor positions, planned and unplanned stoppages, mechanical changeovers and crewing. When users specify the relevant products, it incorporates data on routes, liquid changeover times and physical dimensions.  

Diageo can use the model to simulate packaging operations across both glass and can lines. The glass lines can be fed from both returnables and non-returnables, and up to 3 packaging lines can be analysed concurrently. As a result, the model supports ‘cross-overs’ – for when product groups need to use equipment shared between lines.  

The model simulates production schedule scenarios to ascertain facility performance in various conditions. It reports on all aspects of performance, including: 

• Product run duration and batch performance
• Minute-by-minute analysis of all causes of losses
• Equipment utilisation 
• Conveyor utilisation (i.e., accumulation used)
• Operator utilisation (by skill level and type)
• Production target adherence
• 6-loss analysis

To ensure model accuracy, we configured it based on the Runcorn and Belfast facilities and validated the output using current performance. Then, we conducted a series of experiments with various ‘what-if’ scenarios to look at the effects of different production changes.   

The impact 

Multi-million-pound savings from boosting efficiency and avoiding unnecessary capex  

Thanks to Witness, Diageo managers can analyse options for efficiency improvements in a risk-free environment – before starting engineering work or incurring costs. They therefore gain an accurate understanding of the benefits and consequences of process and equipment changes, so they can make evidence-based decisions.  

Within the first 3 months of using the Witness model, Diageo:  

• Achieved a record 80% efficiency – the model demonstrated that a line could perform at 80% efficiency if minor stoppages in 2 particular areas were targeted. Diageo addressed those issues and achieved the 80% efficiency, which was a new record
• Avoided £80,000 of unnecessary capex – the model showed that an additional accumulation option being considered had insufficient impact on performance, so Diageo avoided a costly capital investment
• Validated the business case for a £400,000 shrink wrapper – the model demonstrated the benefit of having an additional shrink wrapper during changeover and run-up, justifying the investment cost
• Saved millions of pounds – by determining that installing a mid-line buffer would have little benefit

Given these cost-related achievements, predictive simulation is now widely recognised as an effective tool within Diageo – and has facilitated better buy-in for plant changes. This led to a new policy: ‘All future line configuration changes will not be sanctioned until their real benefit has been simulated,’ according to Jimmy Hall, Manufacturing Manager at Diageo. 

• Optimising scheduling rules
• Setting realistic daily targets based on a weekly schedule
• Optimising skill levels and resources
• Stress testing new products/systems/rules in a wide variety of conditions, greatly reducing the risk inherent in change
• Challenging operating practices and assumptions to understand how to get the best use out of existing assets

Diageo now uses the Witness model to investigate further performance improvements across all packaging sites. As Doug Nicholls, Technical Director at Diageo, said: ‘This is one of the best investments we have ever made.’