Measure performance
The below two real life examples show how the variability in unimpeded run-times varies greatly on different inter-station links. To improve performance it is important to ascertain why there is a difference and then what can be changed to make the "high variability" link more like the "low variability" link.

To understand performance it is important to understand the causes of run-time variability by comparing the two links. Drivers are obviously interacting with the infrastructure in different ways on the "high variability" link and interacting in a similar fashion on the "low variability" link. It is important to understand why so that driver aids on the track, such as signage, can be improved and driver training targeted to where it will most improve performance. From the knowledge gained at this stage, and comparing more links, basic recommendations can be made. However, to give precise recommendations, it is important to understand how all the drivers of run-time variability interact to ensure, for example, that one problem is not being solved and another created.

Conversion of understanding into modelling and simulation tools
When comparing more links, which show differing levels of run-time variability, there is soon an overload of information. In order to understand the interaction between the drivers or causes of run-time variability the understanding needs to be converted into a model where it can be tested and validated, by comparing the infrastructure characteristics with both modelled run-times and run-time variability and actual observed run-times and run-time variability.

Convert the understanding into improved performance through implementing assistance and incentives to those who can impact on performance:

• The output of a Run-time Variability Model will enable the derivation of accurate rules and guidance for:
  • Where signage is required and where it should be positioned. This can then be included within the PPP performance regime as part of existing proposed standards changes reducing both London Underground and Infraco risks;
  • Simulation parameters to measure changes in infrastructure performance which improve PPP incentives whilst simultaneously reducing London Underground and Infraco risks;
  • Where to target driver training and what the training needs to include;
  • The prediction of improved performance from initiatives and hence the ensuing passenger benefits. This improves the derivation of cost benefit ratios and facilitates the efficient prioritisation of works.

These four points are described in more detail below in terms of the key actions to improve performance:

• Knowledge transfer through training. Improved training initiatives can usually be implemented in short time scales, relative to developing new tools and making infrastructure changes. To minimise the delay in performance benefits being reaped the knowledge gained from the Run-time Variability Model would feed into training initiatives to target links with the highest variability and target the causes of the variability. Examples would include training drivers to observe unsigned changes to the permitted speed profile and assisting drivers to identify the point at which to begin braking for a station where these have significant impacts on run-time variability;
• Tools to assist individuals to understand and measure their individual incremental impact on overall performance. The understanding of run-time variation gained would allow meaningful link specific run-time and run-time variability targets based on knowledge rather than guesswork. Combined with performance monitoring, using the systems in place such as the Management Information System (MIS), the effectiveness of training schemes could be compared and improved through identifying areas of best-practice and promoting best-practice at other locations. The overall impact of improvements to run-time variability in total passenger hours (i.e. the passenger benefits) could also be measured through models such as the Train Service Model (a network simulator) thereby allowing schemes to be ranked and prioritised according to the cost benefit ratio;
• Incentives to reward those improving performance and disincentives to those contributing to a worsening of performance. Infrastructure and signage are the responsibility of the Infracos (the private companies responsible for the infrastructure) and performance is incentivised through the PPP performance regime. The Run-time Variability Model will identify the differences between the PPP performance regime run-times and day to day run-times observed in service. This will then enable the improvement of the PPP performance regime in the following manner:
• Set run-time simulation parameters which more accurately reflect in service operations reducing London Underground and Infraco risks;
• Define accurate and precise passenger benefit maximising standards in relation to the positioning and frequency of signage.