Site Search

Train Maintenance Simulations.

Using simulations to improve operating efficiency of maintenance sites.

Ensuring that equipment is correctly maintained is key in making sure that a system is safe and as reliable as possible. Irrespective of which maintenance philosophy a facility adheres to it is important to know that the repair and maintenance facilities used to keep the equipment in good condition are also running as efficiently as possible.

Overall image of train maintenance simulation system

Virtually every maintenance schedule that exists will include some amount of idle time for staff and/or resources and profitable organisations try to keep this wasted time to a minimum. This is where simulation can come into its own, simulation systems can be used to keep track of optimization figures for all staff, resources and buildings and run experiments to find the optimal configuration for a given process. Further experiments could show key factors such as choke points or required stock levels.

The simulation displayed in the adjoining images and video clips was designed to show the operation of a busy train maintenance facility. The simulation displays many of the features of simulation that could be of benefit in this scenario, ranging from variable arrival rates and stock rooms capacities to almost limitless statistical outputs. The model includes the following locations (each location is highlighted by the colours listed):

Also included are 5 different store rooms, colored according to the maintenance buildings they supply.

Description of the Simulation System.

Trains enter the system with arrival times based upon a normal distribution (Alternatively arrivals could be based on a predefined schedule) and travel to the inspection bay, once there the train is inspected and any issues are diagnosed. The inspection process is controlled by a probabilistic approach, with each type of maintenance having different probabilities of being required. Using real world data it would be possible to tune these probability distributions to match the current situation ensuring that the simulation accurately models reality.

Close up of main maintenance area The locomotive will now move between the 6 maintenance/cleaning bays as required (visiting only those deemed necessary by the inspection process), entering each area of the depot when space becomes available and returning to the storage area when no maintenance bay is available. For example a train that needs to visit both the light maintenance bay and the locomotive shop will enter the light maintenance bay first if the locomotive bay has no spare capacity and vice versa, exceptions to this rule include situations such as requiring a respray, which will only be performed at the end of the maintenance procedure.

Each maintenance/cleaning bay will use some amount of disposable supplies or parts during the repair work, these are drawn from various stock rooms throughout the model. The quantity of stock used on a given train varies, and is fully controllable within the environment. Each stock room receives a parts/supplies delivery daily and by varying capacities of the stock rooms it is possible to ensure that the stock rooms never completely empty.

Once all maintenance is completed trains will wait for the tracks to be clear before leaving the system.

 

Demonstrational Videos.

Please take the time to view the following demonstrational videos displaying parts of the simulation described above. The simulation displayed is not related in any way to the background image on which the simulation is run, the image is included to make the model more user friendly and easer to follow.

The video clip above demonstrates a normal run of the simulation, the figures along the top display current values from within the model. The left hand set of figures displays remaining capacity at the various locations and the right hand side displays the remaining stock levels.

 

The use of sophisticated routing conditions allows the trains displayed in the model to visit maintenance bays in an intelligent way. Trains will route to the least utilised maintenance area that is included within its individual maintenance schedule. The total length of the above clip represents approximately 1 hour worth of real time which is timed to coincide with a delivery of locomotive parts.

 

Statistical Outputs.

The real strength in constructing a realistic simulation of a system lies with the statistical outputs that can be derived from the model. The graphical elements of the model are supported by data that is collected each and every time the simulation is run. almost any numerical data used within a simulation can be altered almost instantly, allowing for differing scenarios to be played out concurrently and the results directly compared. This means it is possible to test out ideas within the system as well as creating experiments that the system can run itself.

This demonstration model is currently set up to include 4 scenarios that compare the results of the following:

Average train utilisation charts Within minutes the system can check these different scenarios against one another and supply both raw data and usefully charted information to the user (Nearby images display charts for the four scenarios, left clicking on an image will load a larger version in a new window).

Using these four options as examples the program quickly computed that increasing the light parts store will have a beneficial effect, drastically improving service times (in fact this simple step saved an average of 2.5 hours per train!). This beneficial effect can be seen from the improved utilization figures for the trains that pass through the system.

Increasing the amount of visiting trains by a significant amount has the unfortunate effect of overloading the system and causing tailbacks. This in turn slows down throughput and results in many trains being turned away due to insufficient capacity within the yard.

Reducing the speed limit would at first glance appear to be a reasonable course of action, after all, the simulation does show that the amount of time trains spend traveling is a small proportion of the total time. The simulation quickly shows that, as far as service times are concerned, this is a very bad plan with an average of 50% of trains having to be turned away due to insufficient capacity. A different statistic measured by the simulation reveals the cause, in the original run there was a train in motion approximately 72% of the time, with the new speed limit this increases to 98%. This means that the chances of a train being able to move to a new location when it is ready is only 1 in 20 and it is this fact that cause the delays within the system.

Charts displaying the contents of the store rooms with respect to time

Scenarios are very handy when a new idea has been suggested and it needs to be tested before implementation. The above scenarios would lead quickly to the implementation of an upgrade to the light parts store room while discounting the other options. The alternative is to use the simulation system to automatically provide an optimal solution by running a series of experiments. Experiments work by being supplied with a series of modifications with associated costs or savings, and an optimization objective.

Possible modifications could include improving service times in certain maintenance bays, modifying stock levels or improving the track systems to allow more traffic around the yard. An example of a optimization objective would be to increase throughput by the maximum amount without investing more then a set amount of money.

Given the parameters and the optimization objective the simulation software will run though every single possible combination of possibilities (normally more then once each) until it finds the optimal solution with its associated cost and related benefit/improvement.