An international review comparing rolling stock maintenance and performance at passenger operators around the world found a wide variance in cost-efficiency, reliability and availability. An exchange of good practice offers the potential for significant gains in cost and reliability


Heiner Bente is Managing Director of BSL Management Consultants, and Adriaan Roeleveld is Business Development Director at NedTrain

IN RECENT YEARS, Netherlands Railways' passenger business NS Reizigers and rolling stock maintenance division NedTrain have faced widespread dissatisfaction among passengers and the general public over rail's quality of service and level of performance. With competitive pressure on costs increasing, the two divisions launched an international study to benchmark their performance against peer railways and identify good practice which could be adopted.

BSL Management Consultants was brought in to support NS in developing and conducting the study, which focused on two main areas:

  • the cost of maintaining rolling stock, ranging from daily cleaning and servicing to regular light and intermittent heavy maintenance;
  • rolling stock performance, with particular emphasis on reliability, failure rates and availability.

To provide a peer group, NS approached 12 passenger operators around the world with comparable traffic levels, who agreed to co-operate fully with the study. Most are from Europe, with some from North America. The first full set of results was produced in mid-2005. Individual performance figures are kept anonymous, except that each participant knows their own results and their relative position within the group.

The initial analysis unveiled some interesting findings in terms of cost and performance, but also in terms of interpretation. This article aims to summarise our general conclusions for the benefit of other fleet owners, operators and maintainers - and maybe for manufacturers as well.

Characteristics of the sample fleet

The study covered around 8000 vehicles, of about 50 different types. The majority were electric multiple-units, but there were also some DMU cars and loco-hauled trainsets.

Fleet operating characteristics vary considerably, depending upon traffic patterns and operational requirements. For example, Fig 1 relates annual performance to the average distance between stops. For commuter operations, inter-station distances are typically less than 5 km, which leads to very short acceleration and braking cycles. Mechanical wear is higher, and this impacts on maintenance costs.

On the other hand, utilisation can vary dramatically, from under 100000 km to well over 200000 km per vehicle per year. Apart from the commercial implications of having the capital cost per train-km varying by a factor of three, the operational performance also impacts on unit maintenance cost per vehicle-km. Given that there is a base level of service and maintenance activities for each vehicle which is effectively a fixed cost, there is a strong degressive function when expressing total maintenance costs as unit cost per km.

For benchmarking purposes, we recognised that comparing maintenance costs at 'face value', without considering the operational circumstances, would not provide meaningful results. The key question was how much peer group operators would spend to maintain their rolling stock if they had to operate under the same circumstances.

To this end, we developed a normalisation method that reflects the cost impact of different operating characteristics and provides a suitable translation tool (Fig 2). The normalisation functions were derived from quantitative analysis of existing fleet data. Different price levels, exchange rates and purchasing powers were normalised by using cost data published by OECD. Using these normalisation functions, the maintenance costs for each subset of the fleet sample could be translated into the Dutch reference situation for comparison.

Maintenance costs

Fig 3 shows the normalised maintenance costs for EMUs and DMUs per vehicle-km. As well as the fundamental differences between types of train, there are clear variances between participants within the different train types.

Typically, the unit cost for DMUs turned out to be twice as high as for EMUs. With loco-hauled trains, traction maintenance is approximately twice as expensive as that for coaches. This applies to both the average unit costs across the sample and for those figures representing best practice.

Individual results vary significantly compared to the average. This could be due to the relative ease of maintainability for the different train fleets or to the overall cost-efficiency within the maintenance organisation. For both types of multiple-unit, a cautiously-determined 'good practice' result appears to be 40% more cost-effective than the average. Similar findings apply to locomotives and coaching stock sets.

NS has found it very helpful to derive its own cost targets from this analysis of good practice. Discussions within the peer group have helped to identify in more detail where lessons can learned.

Tracking performance

Rolling stock reliability is measured very differently by different operators. A relatively common basis is to record train-affecting failures that cause delays above a certain time threshold, typically 5 min. Since the train compositions, in terms of traction units and the number of vehicles, can vary widely, and a failure in a single vehicle can 'ground' the entire set, a meaningful benchmark needs to look at the mean distance between failures in terms of both vehicle km and train-km.

EMU reliability rates (Fig 4) show even greater variations than those for maintenance costs. Although in one case MDBF exceeds 100000 train-km, most results in the sample fall clearly below 50000 train-km. Looking at the base reliability of individual EMU cars, good-practice MDBF rates are typically twice as good as the average. By contrast, MDBF figures for individual DMU vehicles are on average only half as good as the comparable EMU cars.

As with the maintenance cost, interpretation of the initial results has to differentiate between three possible causes, which would typically be mixed in different proportions in practice:

  • deficiencies in the reliability of rolling stock components and subsystems;
  • flawed execution of maintenance and inspection;
  • non-optimal maintenance and inspection intervals.

Given the wide variation in the results and what appears to be a relatively poor overall level of reliability, performance appears to be an area where much work remains to be done.

For its part, NS has come to the conclusion that an extra effort is needed to raise performance levels further on its densely-timetabled and intensively-used commuter-style network. In the longer term, this is likely to include a complete redesign of the whole process and train servicing logistics interface between NS Reizigers as the operator and NedTrain as service provider.

For both owners and operators, fleet availability is as important as reliability. Again, the measuring and reporting of availability was found to vary between members of the peer group.

A pragmatic definition of availability is the number of vehicles ready for use on peak revenue services. Non-availability can be categorised in terms of:

  • the number or percentage of vehicles needed as an 'operational reserve' for extraordinary traffic situations;
  • the number or percentage of vehicles absorbed for planned and unplanned maintenance, including ancillary activities - the so-called 'workshop reserve'.

Given the background to the benchmarking project, at this stage NS has concentrated on tackling the workshop reserve issue, rather than the operational reserve.

Fig 5 shows that average workshop reserves are 8% for DMUs and 10% for EMUs. Once again the individual figures are strikingly different, particularly for EMUs which range from less than 5% to 15%.

Interpretation and deeper analysis suggest there are several factors involved. Issues of reliability impact on the levels of unplanned maintenance work, and there is also a difference between preventive and corrective maintenance strategies.

Some operators have strategic service concepts that build maintenance policy around traffic requirements and prioritise availability at key times.

The cost of capital represented by the excess fleet in the workshop reserves is sometimes hidden, neither explicitly analysed nor managed. Based on typical figures gained from analysis of the peer group, the capital costs of holding a high workshop reserve can add as much as 50% on top of the unit maintenance cost for a given fleet.

From the owners' and operators' perspective, it is worth managing this cost as effectively as the maintenance process itself. For the maintainer, there should be scope to charge somewhat higher maintenance prices for adopting a strategy that reduces significantly the proportion of the fleet out of service.

The optimum trade-off

The key results from the benchmarking project are summarised in Fig 6, which combines both the maintenance cost and performance factors.

A clustering of results for all fleet subsets indicates that maintenance cost efficiency and reliability are not necessarily concurring objectives, and might not be achievable at the same time. Again, the bulk of the fleets are grouped around the average cost and performance levels, with a number of good-practice subsets achieving higher reliability at fully-competitive cost levels. Some of the best performers achieve twice the reliability at a cost level somewhat lower than the average.

A more detailed analysis found that as a broad trend those fleet subsets with high reliability are achieving the best availability, and more importantly that the normalised unit maintenance cost decreases with higher availability.

This second finding may seem at first glance to be somewhat counter-intuitive. However, it is very plausible, considering that poor reliability necessitates larger workshop reserves for unplanned work, distorts the work-flows and involves corrective interventions that can be far more costly than regular maintenance. It is also supported by the fact that we could find no correlation between maintenance cost and average fleet age.

On the other hand, it is also striking that different fleet subsets maintained by the same organisation sometimes show major variations in unit maintenance costs, even where a consistent level of cost efficiency can be presumed.

Learning the right lessons

The benchmarking analysis has given us a good framework to compare rolling stock maintenance costs and performance, accounting for different operational circumstances to provide a good assessment of relative cost-efficiency. The results from our initial studies indicate substantial differences in efficiency, reliability and availability, which apply to different vehicle types and organisational structures.

The analysis has identified several examples of good practice within the peer group. For the second phase of the project, NS Reizigers and NedTrain are hoping to establish a deeper exchange of information with the best-performing maintenance service providers.

The three main areas that have been identified for closer attention are:

  • the design and construction of rolling stock, designing for maintainability with fault monitoring and advanced diagnostic systems, and improving vehicle reliability;
  • operational processes at the interface between revenue operations and maintenance, such as logistics, workflow processes and defect handling;
  • maintenance processes, including the adoption of predictive and condition-based preventive maintenance, and improving throughput times to boost availability.

NS Reizigers expects that further significant gains can be made both in terms of cost and reliability, which we believe will benefit its train operations, and ultimately rail passengers in the Netherlands as well.

  • Picture caption: Fig 1. Performance data from members of the peer group was normalised to reflect the different operating conditions. Annual utilisation varied from 50000 to well over 200000 km per vehicle per year
  • Picture caption: Fig 2. The dominant cost drivers which had to be normalised were the average distance between stops and the average annual vehicle utilisation
  • Picture caption: Fig 3. Maintenance costs for the peer group varied by a factor of two for multiple-unit vehicles and by a factor of three for loco-hauled stock, but in each case it was possible to establish a band indicating good practice
  • Picture caption: Fig 4. EMU reliability can be expressed in terms of trainsets or individual vehicles, based on the average number of vehicles per train, but in both cases there is a factor of more than 10 between the best and worst performers after normalisation
  • Picture caption: Fig 5. Multiple-unit availability ranges from 84% to 98%. EMUs produced the best availability figures amongst the peer group, but the average was higher amongst the DMU operators
  • Picture caption: Fig 6. Benchmarking normalised maintenance costs against the mean distance between failures revealed a group of well-performing EMUs achieving significantly better reliability than the good practice band

La comparaison identifie les bonnes méthodes d'entretien du matériel roulant

Un tour d'horizon international comparant l'entretien et le fonctionnement du matériel roulant au Pays-Bas avec un groupe d'opérateurs voyageurs de référence dans le monde a débouché sur une grande variété de rentabilité, fiabilité et disponibilité. Utilisant un cadre analytique afin de réfléter les différentes conditions d'exploitation a révélé les meilleurs exécutants dans des domaines de co-t et de l'exploitation. Les NS cherchent maintenant à réaliser des gains significatifs à la fois en matière de co-ts et de fiabilité grâce à un échange de bonnes pratiques

Benchmarking identifiziert güte Praktiken beim Rollmaterialunterhalt

Eine internationale Review, welche den Rollmaterialunterhalt und dessen Leistungsfähigkeit in den Niederlanden mit vergleichbaren Personenverkehrsbetreibern weltweit vergleicht, kommt zur Erkenntnis, dass eine breite Spanne in Wirtschaftlichkeit, Zuverlässigkeit und Verfügbarkeit festzustellen ist. Mit einem analytischen Framework welches verschiedene Betriebsbedingungen bei den Vergleichsbesten in Bereichen wie Kosten und Betriebe, sucht NS nun mit dem Austausch von güten Praktiken wesentliche Fortschritte sowohl bei den Kosten, wie auch bei der Zuverlässigkeit zu erzielen

An lisis comparativo para identificar las buenas pr cticas en el mantenimiento del material rodante

Una revisión internacional llevada a cabo en Holanda, en la que se compararon técnicas de mantenimiento de material rodante con un grupo de empresas de transporte de pasajeros de todo el mundo, halló grandes variaciones en la rentabilidad, fiabilidad y disponibilidad. Mediante el uso de un marco analítico en el que se reflejaban las distintas condiciones de operación, se identificaron las empresas con mejores pr cticas en reas tales como los costes y las operaciones. NS est buscando en la actualidad formas de lograr ganancias significativas tanto en costes como en fiabilidad en un programa de intercambio de buenas pr cticas