Relayr does not just work with partners on production lines but also on the planning processes behind them. A large mechanical engineering company that produced bespoke, non-standard components for wastewater solutions needed assistance in finding a system to estimate the time it takes to produce custom parts. They were manually predicting the time using best guesses. The mean average percentage error (MAPE) of the human estimation was as high as 26%. This inaccuracy caused production slowdowns as there were longer wait times from one factory station to the next. The company provided relayr with a dataset for a common component type, including details about length, width, etc., and the actual production time it took to make them.

Using this basic information, a relayr machine-learning algorithm created a prediction template that could be used on other parts types. This reduced the mean error values to 14.6%. The business outcomes included increased accuracy due to the visualization of real-time data rather than the previous system where estimates were used to plan production cycles. This change in turn helped to increase productivity as it gave workers more concrete goals. The improvements also reduced labor costs as the time-consuming ‘guesstimation’ was eliminated from the planning process.

The relayr team was called in to improve the performance of a large production line for a major international manufacturer and distributor of car suspension parts. The production line in question was actually a number of machines and was key line at the company’s main manufacturing base. After the initial implementation phase where the relayr team was tasked with improving line performance and reducing losses, the visualized data revealed part of the system that was producing up to 100 faulty parts per shift (around $4k per shift). The data also revealed critical bottle necks at a gas filling machine on part of the line. Both of these issues could then be dealt with appropriately.

In the post-implementation phase, the team then worked on helping the manufacturer to predict and therefore speed up changeover times on the line. They developed a stock- taking widget that showed real-time critical buffers and prepared shop floor crews for the impending line change with SMS alerts. The immediate outcomes were enhanced by the newly installed predictive maintenance alerts and anomaly detection systems that were left in place. The company now had remote, centralized control over the entire production line.

Relayr was called in by a major parts manufacturer and servicing agent for bottling plants. They were tasked with improving OEE at all the plants under contract.
Production lines had capacities of up to 1.2m bottles using machines with an average age of 20 years. As well as increased productivity and lower maintenance costs (the management were under budgetary constraints), the team were asked to reduce unplanned downtime, which cost up to $35,000 per/hour, advance OEE values and improve product quality.

Relayr retrofitted the existing machinery with sensors to collect previously unseen data that was then presented on customized dashboards. The data revealed key parameters that could help the operators to improve product quality and minimize waste. The anomaly detection features identified three functions that were responsible for 90% of their plants’ downtime - filtering, packaging, and labeling.

Due to a 20% increase in machine uptime, the client achieved an 11% performance improvement overall. The 8% quality improvement and 30% reduction in maintenance costs led to a 2% increase in operating profit margin.