Uncovering the once undetectable with Point Machine Health

Punctual trains depend on the reliable operation of point machines whose failures are rather frequent and cause about 60% of S&C-related delays. KONUX Switch is combining track and signalling insights to enable next level point machine health monitoring, making it possible to reduce switch failures by 50%., 05.10.2023

The efficient operation of point machines plays a critical role in maintaining the availability of the railway infrastructure and facilitating the movement of trains to their intended destinations. As the first end-to-end predictive maintenance solution for railway switches, KONUX Switch is now combining track and signalling insights to enable point machine health monitoring, ultimately enhancing higher operational quality in managing Switches and Crossings (S&C).

Understanding the challenges of current solutions

Point machines, also known as switch machines or point operating equipment (POE), are devices used to control the movement of rail switches. Their primary function is to align the points or rails of a switch to the desired position, allowing trains to continue on a specific track or to divert onto a different one. When the switch cannot be securely moved or locked, the track is blocked to ensure safety, however, this results in delays to train operations. While repairing a point machine takes considerably less time than repairing a switch rail or a crossing (also known as a frog), point machine failures occur much more frequently, accounting for approximately 60% of S&C-related failures and the resulting delays.

Condition-based monitoring for point machines has been commercially available for two decades, primarily capturing electric or force measurements during a switch movement. These systems are well established and widely used by infrastructure managers. However, the expected impact on reducing failures and the number of stopped trains falls far short of expectations due to several reasons.

  • Firstly, most point machine monitoring systems rely on simple statistical thresholds that cannot effectively detect complex fault modes.
  • Moreover, these solutions are prone to human error due to reliance on manually set reference values and are unable to adapt to dynamic conditions such as temperature and rainfall.
  • Finally, existing solutions typically provide only data instead of actionable insights and do not consider other critical switch components. This lack of root-cause analysis often results in recurring failures.

All of the above makes it very challenging for users to plan and execute the right activities to ensure optimal interventions for point machines.

Introducing Point Machine Health for KONUX Switch

KONUX Point Machine Health detects and diagnoses underlying faults of different subcomponents using electric current data, alerts users to potential failures while factoring in weather data and track condition, and assesses the effectiveness of maintenance actions. This holistic approach helps our users identify root causes of failures and optimize maintenance interventions, making it possible to reduce switch failures by 50%.

In addition, our solution can be integrated with your existing data sources to upgrade your current monitoring systems with enhanced capabilities. Alternatively, we are able to supply proven electric current monitoring hardware in locations where it does not already exist.

KONUX Switch

Detect and diagnose. Point machine failures often result from a combination of faults which manifest themselves in the measured data. Therefore, early detection of evolving faults is crucial for timely targeted maintenance before a failure occurs. Using healthy throws in the historical data as a baseline, our algorithms are trained to diagnose the most common fault types (from locking failures to failing motor components, and broken drive rods).

When combined with track data collected by our KONUX Switch IIoT devices, our point machine models are then able to identify switches with a high risk of sudden failures that would not normally be detectable using electric current data. Additionally, they can help uncover the root cause of the fault and identify whether the defect is caused by the underlying track. This holistic approach enables a better understanding of your switch health both above and below the rails.

In the example below, our system detected and diagnosed (with high confidence) an intermittent locking fault for a point machine in early April, ten weeks before its eventual failure in mid-June. Acting on this insight would have allowed for a more timely intervention that would have prevented costly delays.

Example 1: Detect and diagnose

Alert and assess. While not every fault results in a failure, informing users about detected faults allows them to investigate the event and take necessary actions if required. When significant events are detected by our system, alerts are sent to users with context, including track-related insights, if available. Furthermore, these alerts are adjusted for weather variations to ensure that we do not send false alarms. This means that our users only receive relevant alerts that truly matter. Additionally, when a maintenance activity is conducted, our system assesses its effectiveness and identifies if the root cause has been resolved.

In the example below, the point machine was readjusted following a failure to lock in mid-June. This maintenance activity was assessed to be effective since it eliminated the underlying intermittent fault.

Example 2: Alert and assess

KONUX Point Machine Health is designed to assist our users by detecting, diagnosing, alerting, and assessing faults related to point machines and their associated switch. By providing timely insights into when, where, and why an action is needed, it enables users to ultimately avoid repeating failures, reduce asset lifecycle costs, and increase network reliability.

Moreover, our embedded feedback channel allows users to provide additional context and ground truth, facilitating continuous learning and improvement of our models. The data visualization builds trust and preserves knowledge that can be shared across the customer organization, promoting cross-organizational learning and maintaining critical maintenance knowledge.

What’s next?

We are actively working on integrating additional existing data sources, such as track geometry, point heating and downward-facing cameras, to detect even more faults that were previously considered undetectable and further enhance our ability to identify root causes. Ultimately, we would like to provide a comprehensive one-stop solution for all your switch-related needs.

. . .

Would like to learn more about our solutions? Get in touch here.