TIPOS DE PRODUCTOS QUE FABRICA LA INDUSTRIA PETROQUIMICA

During approximately the same period as the D 23 studies done in Europe, the Asso- ciation of American Railroads, AAR, undertook a series of field measurements of the dynamic amplification factor on American railway bridges which was subsequently reported by the American Railway Engineering Association, AREA,.

The article (Byers, 1970) uses the field measurements of the AREA to make sugges- tions for the distributions to be used in a probabilistic code. The articles provides a useful summary of the AREA survey describing types of bridges speeds and impact factors. A normal distribution was found to be the best fit to the field data and in contrast to (Tobias, 1994) found the log-normal distribution to poorly represent the data. This paper was based on the results using diesel locomotives. It is somewhat unclear from this article and that of (Tobias, 1994) whether the tests were carried out using service trains or just locomotives. In this article it is stated that the AREA test program involved 37 girder spans with some 1800 test runs using trains with diesel locomotives. Of the bridges in the span range 10–20 m, eight were open deck and seven, ballast deck. The ballast deck bridges were of concrete, timber and steel plate. However, the article does not describe the actual bridges in this 10–20 m range.

For 10–20 m unballasted bridges and speeds in excess of 40 mph (64 km/h) a normal distribution was fitted to the data with a mean value of 24% and a standard deviation of 15%. There is a very interesting table in the said article that lists for the two categories of bridges (ballast or open deck) the mean impact factor, the standard deviation, the number of tests for each train speed category.

The article also stated that, not surprisingly, ballasted bridges had a lower impact factor than comparable open deck bridges, this phenomena is also noted in a com- parable study done in Europe (Specialists’ Committee D 23, 1970c). The article also provides a good description of the problems of axle loads and dynamics on bridges. For example it comments on varying sources such as high tolerances in the fabrication of goods wagons and locomotives for springs and components, wear on wheels constantly changing the dynamics, track irregularities both on the bridge and on the approach to the bridge as a major factor to the resulting impact factor. Also mentioned is the effect of run-in and run-out i.e. the taking up of slack due to braking and acceleration of the train between wagons. This causes pitching of the

wagons and hence a redistribution of the load to adjoining wagons and boogies via the couplings. I am not sure how relevant this is for European wagons as they do not have the same coupling system as the US.

The article (Sanders and Munse, 1969) discusses the distribution of axle loads to bridge floor systems. This is not the subject under discussion, however, it does refer to the AREA tests and it would appear that the tests solely used information related to locomotives in service trains and not the entire train. It also states that locomotive axle weights could vary as much as 20% compared to the specifications of the manufacturer. It should be remembered, however, that this is based on old manufacturing procedures where a vast majority of the locomotives were steam locomotives.

Articles (Tobias et al., 1996; Tobias and Foutch, 1997) are both very interesting, they present work carried out in the US on statistical analysis of measured axle loads using the same technique, although in a simpler form, as that used for the data collection of this thesis, see Chapter 7. The article (Tobias et al., 1996) mostly presents studies of the axle loads. It presents information obtained from field studies of five instrumented bridges all steel riveted.

The other article (Tobias and Foutch, 1997) has a description of the distributions used for the loading and a description of some Monte-Carlo simulations that they have used for the assessment of the remaining service life of steel railway bridges. It uses reliability theory for this purpose. The original dimensioning load for these bridges, built after the war, were the steam locomotives and not the trailing weights from the wagons which is a comparable situation to the Swedish traffic loads for this period. The article also refers to the study on the impact factor done by the AAR. The article finds that the best distribution to describe the impact factor is log-normal. Fig. 5 of this article shows that for many of these steel bridges the mean measured DAF was approximately 1.1. These values can be compared with a summary (Hermansen, 1998) done on the ORE report (Specialists’ Committee D 23, 1964b) which states that for prestressed concrete bridges the mean DAF was approximately 1.1 and maximum noted was 1.3.

The Ph.D. thesis (Tobias, 1994) studies the fatigue evaluation of riveted steel rail- way bridges. The thesis presents Monte-Carlo traffic models for evaluating the load effects in riveted railway bridges for American conditions. It uses results of weigh- in-motion studies, locomotive and wagon types and train frequencies and configu- rations. In the thesis, the dynamic effect is found to closely follow a log-normal distribution. In Figure 2.1, which is redrawn from (Tobias, 1994), the fitted dis- tributions are shown for the 9.1–18.3 m ballasted decks for different speed ranges. The properties of the fitted log-normal distributions are shown in the thesis for both open deck and ballasted bridges, for differing ranges of span and for the ranges of train speed shown in Figure 2.1.

0 5 10 15 20 25 30 35 40 0 0.05 0.1 0.15 0.2 0.25 Dynamic Effect [%] Probability < 32 km/h 32 - 64 km/h 64 - 96 km/h 96 - 128 km/h 128 - 160 km/h

Figure 2.1: Probability vs. dynamic effect for fitted log-normal distribution, for bal- lasted decks with spans between 9.1 and 18.3 metres. Redrawn from (Tobias, 1994).