Forecasting of Future Tie Requirements by Computer Modeling
Date
1993-07
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Railway Tie Association
Abstract
As railroads watch their costs and expenses more closely the need to more effectively plan their track maintenance activities takes on an increasing importance. This is particularlyimportant for the high capital cost items such as cross-ties which represents a sizable portion of any railway's maintenance of way budget. As these costs continue to increase, the ability to properly plan and execute track maintenance programs in an efficient and cost-effective manner becomes of even greater significance. Through the use of engineering based predictive equations, in conjunction with railroads historical and current tie condition data predict, (i.e. their tie data base) it is possible to accurately predict, on an annual basis, how many ties will have to be replaced, not just next year, but five, ten, and even 20 years into the future. This information, in turn, allows for the accurate forecasting of budgets and budgetary requirements. A key aspect in the implementation of a cross-tie forecasting system is the development of effective failure (or life) relationships for the ties. These relationships are essential to the forecasting of failure, and hence life, from the current condition and historical (installation) data base Wood tie life has been the subject of extensive modeling attempts using various modeling techniques, to include; mechanism models, empirical models, and statistical forecasting models. However, one major difference between tie failure models and models for other track components such as rail, is that ties even when installed at the same time under identical operating conditions, do not all fail at once. Rather, thereis a statistical distribution of tie failure and hence replacement, around an "average" tie life, as shown in Figure 1 for wood ties with cut spike fasteners. Such statistical distribution curves have been developed by the USDA Forest products Laboratory Association of American Railroads. Thus there are, in fact, two possible tie life modeling approaches, an "average" tie life modeling approach which determines an "average" tie life for a given set of conditions, and a statistical tie life approach, which predicts the actual number of ties failed (and thus replaced) each year. Both are needed in order to accurately forecast future tie requirements, particularly if the history of previous tie insertions has not been uniform (which is almost always the case). It should be noted here that the traditional railroad approach to determining tie life is the system average tie life approach, where the total number of tie replacements system wide or nationwide are added, together with the associated trackage and a single value for system tie life calculated. This approach, however, is extremely misleading, since it gives a system average that is not representative of any actual conditions.
In fact, by including the very low density lines, an erroneously "high" average tie life can be obtained. One recent approach developed by ZETA-TECH Associates of Cherry Hill, New Jersey combines the above two techniques. By utilizing a combination of the statistical distribution approach (based on the Forest Products Laboratory failure distribution curves) and an engineering based "average" tie life equation, ZETA-TECH has developed a technique to predict future annual tie replacement requirements based on a railroad's actual tie replacement history, and trackage distribution, specifically curvature, tonnage, and geographic (environmental) distribution. In this modeling analysis, the railroad is divided into a matrix of parameters, specifically annual tonnage (MGT) and curvature. For each element in the matrix, the total miles of track is obtained from railroad supplied data. Given the strong sensitivity of tie life to curvature and tonnage, this represents an essential step in determining the railroad's tie requirements. Then for each of the above annual tonnage-curvature combinations, an average wood tie life was calculated, based on a ZETA-TECH Associates, Inc. model for tie life, calibrated to industry average behavior. Using this distribution of trackage by tonnage category and curvature and the corresponding cross-tie life each category, the annual replacement requirements for each segment is calculated on a "steady state" basis. Thishowever, does not take into account the railroad's historical tie installations. These cycles have been found consistently throughout the last several decades (and even longer) and vary as a function of the economic condition of the railroad, the level of deferred and catch-up maintenance, and the philosophy of the railway at any given period of time. Figure 2 illustrates such a history of tie insertions for a 40 year period. Since the tie installation history is clearly not steady state, the above calculated tie replacement requirements must be modified to account for this history and the corresponding levels of high and low tie installations. In order to account for this, the tie failure distribution relationship developed by the U.S. Department of Agriculture Forest Products Laboratory and validated by the Association of American Railroads was used. This curve, which was presented in Figure 1, shows the percentage of ties failing as a function of the average life of the tie. Thus, based on this Figure, 50% of the ties will have railed at the 94% average life point and virtually all ties will fail by the 160% average life point. By using the Forest Product Curve and the average tie life distribution(by curvature-tonnagecategory) it is possible to determine the percentage of ties failed (and thus requiring replacement) as a function of life, and thus years. This is further modified by the fact that most U.S. railroads, replace ties on a cycle basis. By applying these replacement cycles to the actual tie history, a forecast of future tie replacements is obtained. Correcting this forecast for any changes in track mileage and annual tonnage, over the years, allows for the final calculation of future annual tie requirements. Such a forecast is illustrated in Figure 3 through the year 2010.
Because of the large amount of data involved in this kind of analysis, this approach requires computerization. Such a computerized modeling system has been developed by ZETA-TECH Associates, Inc.and has been applied to several Class 1 railroads. By using this type of forecast information, railroad maintenance officers, financial managers, and senior executives can project future (short, medium, and long term) capital requirements and help control future costs and better plan future budgets. In addition, tie suppliers, treaters, equipment manufacturers and other members of the supply industry can use this type of information to better evaluate their future market and define their future production requirements.
Description
Keywords
Track maintenance, Cross-tie
Citation
Zarembski, A. M., “Forecasting of Future Tie Requirements by Computer Modeling”, Crossties, July/August 1993.