Why did a conflict between two services occur?

Category: Operations
Capabilities: Automatic Conflict Detection Standard Delay Calculations Sub-threshold Delay Calculation

Knowing this, Station Manager can review re-occurring excess dwell time.

Knowing this, Timetable Planner can review re-occurring unachievable SRT.

 

JNCTION is currently providing this capability to LNER via the DST tool, by analysis train movements data from TRUST.

 

Difficulty: Medium 

For a specific location the method can be developed relatively quickly, depending on the complexity of the location. Once set up, insight regarding conflicts can be generated and presented to interested stakeholders for action. 

Data: Mostly Available 

In most cases, JNCTION has obtained and processed the required data to solve this problem. With further data sources we could be able to assign the cause of delays more accurately.  

Scalability: Medium 

The methodology could be replicated across the network easily but would need manual testing and adjustments for location specific nuances.  

Maturity: Low 

JNCTION has completed successful trials in specific locations, producing static conflict reports, but have not scaled to more complex areas. 

 

The Long Read

Why does answering this question help?

Knowing why conflicts occur helps to build improvement plans that reduces the risk of the same conflicts happening again. This helps to reduce delays and the amount of work placed on signallers.

How do we answer this question?

Using train movement data, we attempt to understand each service’s journey up until the point the conflict occurred. 

For each service involved in the conflict we answer: 

  • How late (if at all) was the service? 
  • How late could the service have been without causing a conflict?  
  • For a late service, at which locations did it incur delay and by how much? 
  • For each of those locations, what was the primary cause of the delay? 

Determining the cause of delay will be a complex process. For example, we may identify that two minutes of delay were incurred at station X. We may be able to say confidently that this delay caused the conflict but it will then take further analysis to fully understand the reason for the initial dwell delay. This can be incrementally improved through more advanced algorithms and other data sources. 

Why does answering this question help?

Complimenting the output from “When and where did a conflict between services occur?”, to each conflict identified we’d provide information regarding the causes of the delays to each services, such as excess dwell times or unachievable SRTs. This can then be aggregated over years’ worth of data to estimate the most common causes behind conflicts.