Exposing the real infrastructure cost drivers
INTRO: Why are life-cycle infrastructure costs five times lower in the USA than in the Netherlands, but are 50% to 200% higher in Japan and Hong Kong? Jan Swier* shows how these huge cost differences can be explained, allowing them to be predicted more accurately
OVER THE LAST 10 years, Netherlands Railways has undergone a complete change. So far as infrastructure manager ProRail was concerned, there has been a shift of management focus from technology to contract management, from tasks to processes, from man-hours to costs, and from budgets to operational performance.
In the course of this change, ProRail gained valuable insights into the relationship between life-cycle infrastructure costs and operational performance, and why international benchmarking data proved so valuable in shedding new light on the drivers of those costs.
On the initiative of UIC, a benchmarking study was carried out to compare the costs of infrastructure maintenance, renewals and new projects. The reports published between 1996 and 2002 provided a wealth of information and results (RG 3.03 p130).
One of the most exciting analyses of this benchmarking data, which I implemented at the time, involved plotting the life-cycle costs of railways in some 25 countries spread across three continents against train-km and tonne-km.
Plotting LCC against train-km (Fig 1a) reveals a more or less proportional relationship. It is noteworthy that the three continents form three clearly distinguishable groups that lie more or less in a line. Note that 'Far East' refers to the densely populated and prosperous regions of Hong Kong and Japan where mass transport of passengers dominates. It does not apply to China as a whole, for example, where quite different results could be expected if the data were available.
Plotting the same LCC against gross tonne-km (Fig 1b) generates a totally different picture, and at first I saw no logical explanation. Engineers responsible for rail had always told me that deterioration of the superstructure is caused primarily by the axle load and gross tonnage each section carries.
From my own experience, I know that the rail infrastructure and traffic are very different in the three continents. But what always struck me during my visits was not so much the extent to which railways differed as how much they had in common. Everywhere I went I saw recognisable rail infrastructure, be it track, pointwork, signalling, bridges or tunnels. There were no differences in principle in the type of rail infrastructure. The explanation for the huge discrepancy in costs had to lie elsewhere.
A variance analysis of traffic that the railways in each continent typically carry offered the first clues. Combining this with information in the two graphs slowly but surely transformed speculation into knowledge. This process was further encouraged by a number of relevant questions from managers at ProRail.
What really helped me stay on course in my search for infrastructure cost drivers was a generic management paradigm that I developed.
Every manager's task is to control time, scope, costs and conditions, and so long as they are under control he is in charge. The four elements contain a number of related variables:
Scope embraces functionality, performance and technical quality as well as reliability, availability, maintainability, safety, health and environment (RAMSHE);
Costs are determined by man-hours, materials and machines;
Time for developing and constructing an infrastructure component plus its service life;
Conditions and circumstances that affect the task but cannot be influenced by management.
The four elements comprise a cohesive management framework. If one element changes, its effect on the other three has to be investigated, and the consequences of that investigation accepted. This is a general management principle, and it certainly applies to rail infrastructure.
Management can be compared to keeping a pair of scales in balance, with one side containing money and the other side time, performance and limiting conditions. The manager is able to keep them in balance using two different but complementary methods.
The better known method, if he has less money to spend, is to translate this into less availability, less functionality, lower pay, and so forth. However, there is an alternative policy which is to run the organisation more effectively, thus achieving greater results at less cost. This is possible if his employees perform their tasks better because of greatly improved process management and the use of powerful management tools.
The scales can be considered as a paradigm for management from which all the influences on costs can be derived. This provided me with a checklist, which I used for developing a cost model and the analysis of cost drivers. Thanks to this paradigm and the checklist I have maintained an overview, and remain confident that I have taken practically all the major effects on operational performance into consideration.
The QM4C cost model
On the basis of this insight, I developed the QM4C model for forecasting the LCC maintenance and renewal costs for new and existing lines. QM4C stands for: Q = Quality; M4 = Money, Man-hours, Material, Machines; 4C = Costs, Circumstances, Conditions, Complexity.
The model consists of six modules. The core is a cost template (Module 1) that is linked to a calculation model for running sensitivity analyses (Module 3). The other four modules include norm graphs for determining the service life, overhead costs and a method for determining effective working times, but these are less relevant for the purposes of this article.
Module 1: cost template
This contains infrastructure quantities and cost ratios for maintenance and renewal under average usage conditions on the Dutch rail infrastructure. These are assumed to be 27000 gross tonnes/day passing over each track, 65% of maintenance carried out during the day, and an effective working time of 5h during the day and 4h at night.
Maintenance budgets are calculated by multiplying cost ratios by the infrastructure quantities. These budgets, when added together, should cover total expenditure for the year. The template is indexed for cost inflation and can be brought up to date every year. The cost ratios are therefore realistic and reliable, because they explain the actual maintenance costs in a given year.
The cost template is also used to forecast maintenance costs for existing or new lines. This can be done simply by replacing average quantities for the NS network with quantities for the line to be analysed. This yields a maintenance cost forecast based on the complexity of the line and the average usage circumstances of the NS network.
The cost template contains cost ratios for renewal as well as maintenance. Supporting evidence and historic calculations are available for all those ratios, as a result of which the calculated cost ratios are also realistic and reliable.
Module 3: sensitivity analysis
The output from Module 1 is the input for Module 3. Module 3 uses a norm graph that describes the relationship between maintenance costs and intensity of use. The norm graph has been compiled from eight subsidiary norm graphs, one for each system that makes up a railway line: track, trackbed, level crossings, civil engineering structures, power supplies, signalling, telecommunications and traffic control.
The norm graph is used as the basis for performing the sensitivity analyses. Module 3 contains a number of parameters with which the model can be calibrated for a specific situation. They can be adjusted to best describe a line, a contract area or an entire country. These parameters are: