Business leader Asset Analytics at Arcadis
Asset Analytics in practice: success factors and business value for rail infrastructure
M.W. Zanen MSc (Maarten), Arcadis
N.M. Huijsman MSc (Marnick), Arcadis
R. Dijkhuizen MSc (Roland), Arcadis
The amount of data that is collected about assets is growing exponentially. This applies to technical, financial, geographical and performance data. Despite this wealth of data, the asset manager still has lot of unanswered questions. For example: ‘Can I spend my maintenance budget smarter?’ or ‘How can I reduce the number of malfunctions of my assets?’. With Asset Analytics we can bridge the gap between the data and these questions. In this way, the asset manager can make better business decisions.
In this paper two cases are presented where Asset Analytics is put into practice. With a practical approach of Asset Analytics we want to show what the key factors are in achieving business value.
Case Predictive maintenance of rail switches
This case is about an Asset Analytics project for ASSET Rail, a major rail maintenance contractor in the Netherlands. Together with the client we first assessed the most promising asset to focus on. Based on criteria like potential business value and availability of data the choice was made to focus on preventing disruptions on rail switches. One of the success factors was the close interaction with the maintenance engineers. By combining their input with our own knowledge of failure mechanisms of rail switches a list of critical variables was drafted. This included data like age, type, location and historical disruptions of the switches. But also some less obvious variables like the height of the track and type of soil were added to test on correlation with the disruptions. After thoroughly preparing and cleaning of the data a machine learning model was trained with data up to 2014 to predict the number of failures in 2015.
With the result of the model it was possible to rank the switches on the number of predicted disruptions. The 20% switches with the highest predicted number of disruptions had about 60% of the actual disruptions in 2015. This ranking was then used to optimize the preventive maintenance on these rail switches. The preventive maintenance effort on the 20% worst switches was tripled, were as the 50% best switches, the effort was halved. This optimization is expected to reduce the number of disruptions with about 10%, without increasing the cost of preventive maintenance.
Case Analytics on slippery track conditions
Slippery track conditions are a major problem for rail operations in the autumn period. These problems range from train delays, extra maintenance costs to potential safety issues. Knowing in advance where and when slippery track conditions occur gives the possibility to anticipate. For example by planning the use of Sandite for slippery prevention. Using historical weather, location specific data like the number of trees and slippery track notification data a machine learning model was trained. This model predicts the change of slippery track per day per track.
Maarten Zanen studied Civil Engineering at Delft University of Technology. He joined Arcadis in 2002, where he fulfilled several positions. Since 2005 Maarten has been working as a consultant asset management and analytics for a variety of different rail asset managers (a.o. ProRail, GVB, HTM, RET, RTU, ASSET Rail). Since 2014 Maarten has been leading a team of analysts and consultants in the field of asset analytics in order to reveal valuable insights for both asset owners, asset managers and service providers.
The RailTech Europe 2017 Conference will explore the following three themes:
- Day 1 – 28 March 2017: European Railway Traffic Management System (ERTMS)
- Day 2 – 29 March 2017: Digitalisation in Railways
- Day 3 – 30 March 2017: Maintenance of Rail Infrastructure