Summary of Two Axle Study

To summarise this section the outcome of the results for the required partial safety factor γq is sensitive to the values chosen for the mean value of the dynamic coeffi-

cient, µϕ and the initial properties of the variation of the axle loads within a bogie,

P . The study showed that results were relatively insensitive to the number of bogies

assumed to pass a stretch of line per year, at least for the large numbers that can be expected on a main line route. Also the results show, somewhat surprisingly, that the reliability analysis was relatively insensitive to the assumed coefficient of variation of the dynamic coefficient. However, the study was limited and it is dif- ficult to say if these results are generally applicable, although under the presented assumptions they do provide a good indication as to the significant parameters. Another interesting point that came from this study and is somewhat obvious for these small spans, is that the axle spacings within a bogie is an important parameter together with the perhaps not so obvious parameter, the spacing from the last axle to the end of the buffers. This last spacing seems, in the majority of the studied four- axled freight wagons, to be 1620 mm. However, on a special heavy-haul route in the north of Sweden, there are freight wagons for transporting iron-ore that have much smaller distances between the last axle and the end of the buffer. This allows for a higher concentration of axle loads over a smaller area which can be concentrated towards the centre of the bridge and thus increasing the mid-span moments, which from the bridge’s point of view should be avoided.

Traffic Load Simulations

7.1

Train Data Collection

7.1.1

General

Data was collected from two sites in Sweden that provided information about the loads from each axle, the spacing between axles and the train speed from moving trains as they passed through the measuring site. One of the sites was in the north of Sweden at Notviken and the other more in the middle of the country at Sannahed, see Figure 7.1.

The traffic situation at the two sites are very different. The Notviken site is on a heavy-haul route where the traffic consists almost entirely of iron-ore transports, where the wagons are now weighed before they are allowed to traffic the route, although this was not the case of all the wagons at the time of the data collection. The traffic at Sannahed is of a mixed freight and passenger type and deemed to be more representative of normal traffic conditions, for this reason the main efforts of this thesis has been directed towards this site. The site is on a straight stretch of line where the trains are able to achieve reasonable speeds, at least when considering freight traffic. Speeds of 150 km/h were noted on this line and the average speed of the over 7400 trains analysed was 102 km/h. The allowable axle load at the site was 22.5 tonnes during the analysed measuring period of January–April 2001.

Information about the measuring methods are presented in the next section and is a similar method to that used for measuring wheel weights presented in (Tobias et al., 1996).

It was not possible to receive direct untreated data from the sites and the information available was in the form of text files. The text files contained information on the loading of each axle, the speed of the train and notations for the locomotive and wagon types. From these text files the axle loads were read into a Matlab matrix using a computer program written by the author. The axle spacings were reproduced by first reading in the locomotives and wagon notations and then via a

Sannahed

Notviken

Figure 7.1: A map of Sweden showing the location of the two measuring stations at Notviken and Sannahed.

library of wagon types, their axle spacings and length over buffers, it was possible to recreate the axle spacings of the trains. This reading in of the data and the consequent building of the trains was again done automatically using a computer program written as a Matlab code.

Once the axle loads together with the relevant axle spacings and tags indicating the start and finish of a train had been read from the text files they were converted and stored as Matlab data files. The Matlab data files were then used to simu- late the trains moving over hypothetical simply-supported bridges of differing span, described later in section 7.2 of this chapter.

One noted problem with this method was that in certain cases the measuring sys- tem had not recognised a wagon or locomotive type. This was registered by the measuring system as e.g. 1 AXLES UNDEF or 4 AXLES UNDEF implying one respectively four axles were undefined. This phenomena raises question about the validity of the adjoining wagon not perhaps in the case of a two or four undefined axles as these could be a whole wagon but definitely in the case of a single undefined axle.