Transportation of Dangerous Goods by Rail, Dynamic Risk Assessment and Emergency Planning

I’m really glad to be here with you I’m learning a lot myself and enjoying also from the presentation so far what I’m gonna try to explain is a prototype type of simulation we are building as part of a large research proposal for to create what we call it advanced disaster simulation lab at York University and that’s because York University is one of the only two universities in Canada that has disaster and emergency management program in fact we have the largest in terms of number of undergrad and graduate students and we are creating our first PhD disaster and emergency management in Canada hopefully very soon next year and for that purpose we are hoping to get deep into any logic using this the professional version hopefully next year if we come together again in the next conference we would be able to see if you were successful without funding I hope so and we would have the professional version at that point too and as an introduction I’ll provide you why I’m using and I’m in an allergic user right now a problem that we are looking at this morning and what this demo or prototype model is trying to do it’s basically toward what we call it dynamic risk modeling not only for the purpose of risk assessment but also for disaster preparedness and planning and a bit of the details of this application and illogic application and demo at the end with some notes about what we are going to do next as way of introduction I just would like to mention that I am really glad to see any logic 7 because of so many reason I was using for those of you who are and I’m sure many of you are I was using dynamo that does very software to create dynamic models during eighties some of you might have used not also and I was using arcinfo back in eighties again that was probably the earliest version in the US again very difficult complicated you have to type all those commands and you know learning process was quite something and then just recently moving into agent-based modeling now I had always this dream of being able to combine system dynamic with GIS I I tried and sometimes it works but you had to couple them in a very complicated way and on the other hand once you are talking about agents model agent-based modeling especially in our environment or the discipline I am working it it it helps but it’s not helping a lot if you don’t have good GIS support for it so I had to go back and forth between GIS and agent-based modeling if I didn’t want to use repass or agent analyst with ESRI which were also difficult so when when I have analogic 7.1 and it happened just a week probably late when when I just finished this model but try to convert it to this new version but anyways I mean I’m really glad to see this and that’s why I’m using this is this is in my view the best tool that you can integrate all those things together at one time that’s why I’m using it based on my experience and based on what I have gone through in all those past years so I see this is a natural progression progression of the the the software that is what we have today and I’m hoping to see even more advancement on that so the problem start with this really in 2013 July there was a big event in in

lac-megantic town of Lac Megantic in in Quebec and this is a sir about forty close to 50 people died there was a train derailment carrying crude oil from North Dakota to Eastern Canada and there were lots of problems and I’m not going to explain why this train derailed and what happened in Indy at all but as one of the first experiments very very simple experiments I did with any logic was this that I also posted online if I can run this it’s just to explain that the problem not really the power of this simulation this train had about seventy two or five tank tanks and had four locomotives locomotives deer derailed and it was almost 1:00 a.m. at night and about 40 of these tankers were licked and you had a large fire and explosion there and it took about up to afternoon of the the that day to finish fighting this fire and of course the whole downtown part of this small town was burned to ground basically and now this was my first as I said type of simulation I did with with any logic and I always wanted to have GIS background in it this is now an image but of course if you are able to interact these behaviors and activities emergency response operations with a map that you can work with that would make it much more powerful and so that’s where I’m coming to this so this is just a simulation of what happened it has not much really activities behind it it’s just because I wanted to simulate the the event back to PowerPoint plays yes now every everyday there are lots of trains carrying hazardous materials in Canada and us these are some of the recent events just for trains carrying crude oil not all other hazardous materials and it’s very evident that recently there has been an increase in these events I mean partially is because of the huge volume of oil transportation through the rail and also because of the fact that the pipeline’s are limited they need more investment long-term investment but these are quite easy and accessible so that pressure on railroads has increased and because of that we have more incidents like this these are just some of them black Megantic the point with Lac Megantic was because of the extent of the disaster and the fact that it was one of the it is now one of the world basically largest and severe disaster train derailment disaster soon after that there was lots of discussions in Canada in US and all other countries how we can mitigate this type of train derailment which is of course an important issue and now as I said that the reason behind this is in part because if you look at the even recent statistics in 2008 you were talking about 9,500 this is the average crew tall rail carry annually these are and then you can scale it and see what is the the same amount in 2013 that is you can guess what the change is so we are talking about this and this is going to increase even faster in the next couple of years and so so this is the pressure and this is as I said I was only crude oil as we’re talking also about lots of other materials and because of a lag Megantic trail derailment what happened was that Government of Canada decided to put pressure on rail companies to provide

hazmat information to municipalities that was one of the few actions they took to somehow mitigate this type of events in the future and there were of course this was because of the municipalities pressuring the government and that that happened but my question to to all the municipalities at that point was of okay suppose you’ve got this information what are you going to do with it can for example a small municipality like Lac Megantic or all other smaller towns and even larger cities have the capability to to use this information in an easy way that’s the difference there are tools but none of these emergency managers necessarily have the capacity to to run and analyze them so my solution not an institutional solution or solution that we are in our discipline looking at as kind of looking at it from risk management perspective and trying to identify how we can reduce this risk through identifying those Peaks or spikes in in risk management so we came with this dynamic risk modeling of course most of you I’m sure you you are aware how we can or what type of variables in such type of analysis we have some of them are static some of them are kind of semi dynamic and some of them are real dynamic so we are dealing with a situation we want to model the situation where some of our variables are real dynamic and there are so many examples of why we say this is dynamic this is just one example the weather impacts a lot in in what trains are transporting the very good example was you today we had an interesting example of the changes that happens in pipelines when carrying crude oil also from one end to the end to another end the same here when when they transfer in fact this was in July so that was the peak and one of the reason they say was because of the extent of damage was actually the weather because it was hot summer and the materials and actually changed and during this trip by the time that they reach Lac Megantic they they were quite different in terms of flammability of the materials so this is really dynamic and also you’re talking about dynamics in people population level and all other attributes now for this we use we try to use very simple but still this is in progress I can’t claim that these probability figures that we have reached are are the final numbers that we can work with but I use the literature to find out how we can or what would be the best measurement of train derailment probability for different segments after L that’s what we used and we calculated the probability of derailment based on this past research and you also need to when you do risk assessment you also need to calculate the the exposure in this study we although we consider all assets that are impacted but our focus was mainly on human exposure because that’s the most important key decision-making in this in this regard Government of Canada same as us they have some standards of how much would be the exposure area or impacted area for different types of hazardous materials whether it is they daytime or nighttime this is for major spills because we are not really looking at the smaller spills where you know normal emergency operation can take care of but this these are for larger spoons where distances are usually high so looking at this there are about 600 some hazardous materials carried through this or potential to be carried through rails also and for each of them you have these what they call it on those three last columns you have the eyes isolate impacted area or immediate impact that area for example for for coloring is 10:17 is 500 meters for with a very immediate impact that this area has to be evacuated or people who are in that area might be severely impacted if there is a derailment derailment of this type and so with some yeah and the other numbers are in kilometers so you have two 533 kilometers and 7.9 columns this

is our base for calculating the exposed or impacted area and of course for a smaller spills the downwind is an important factor but for larger spill you have to consider the whole like buffer zone around your derailment event so the overall dynamic risk model would be kind of multiplying your derailment probability by your population at risk at any given point at any given time that’s what we are putting together here and now why we choose Toronto of course this could be applied to any other city and that’s our intention this is not just for Toronto we use we chose to turn to as a prototype because we are in in Toronto and also had better than other for Toronto and in fact because the Toronto is the largest city in Canada the risk is much higher in the city as well and already lots of counselors and others are asking for some change in this this is one of the rail line I mean I’m focusing on right now but it could be different all different right roads inside the city so all the DC councillors are asking this happened of course after this event and now this application I’m working on as I said it includes all those hazmat chemicals it doesn’t mean that all trains carrying this some trains at some points of time they carry this not necessarily large volumes in tankers maybe one tanker over 100 cars is carrying hazmat materials so you have some agents here but these are just some of them you can add as many and in fact based on demonstration I had with the city emergency managers they provided some good feedback I’m going to add some more but it is just a capability you can add as many agents as you want in terms of type here I have for example ambulance services fire services fire station police and police services hospitals you can add long-term care facilities whatever you like of course as you know you cannot but I have shapefiles for all of these I convert them to Latin on long base and there with their attributes into a text file CSV file and then create those agents and now that you can have actually if you have all this information within your OpenStreetMap you can read from there but my test was not showing exact results if you do it by OpenStreetMap for example you you get like 50 schools in Toronto but here I have like 550 schools so that’s why I’m using the actual city information and to calculate the day and night population I had the night population which is assuming that the census data is night population I had the employment number for different districts in the city and I estimate that kind of data on population based on the employment level in the city and different districts of the city and some of these agents are agents related to the actual has hazard like train like plume like particles or fire itself now this is a simple demo of inside for example the population agent this is basically an impact as distance as train enters that particular place so it changes the state of one agent from being normal to impact that night or impacted that day depending on what the material is and this information of course is read through that text file and also pass into this is another example of the Train agent where I have the other again coming from that table that I showed you with lots of columns but some of the key variables here are are these and interactions with other agents through that distance and this information is also passed through dynamic variables to my excel file as a data base and I can transport and use those also in ArcGIS for other activities now this is a typical result

you get for example for petroleum if you have a train carrying petroleum and there is a major leak this would be the day and night impact and as I said we are interested really on those Peaks and if we can lower those pigs we would be able to change the whole dynamics of risk management in a better way so those lines you see our averages for day and night and just by changing some of those upper level pigs you might be able to minimize there is quite significantly and when I showed this intro introductory results to the city emergency after on time agency managers they they said basically we would like love to have this for each ward because our counselors if they see these results and this simulation they would jump into that that’s what they need and that’s what they can convince others to take some actions so this is the structure of the the interface of the the model on one hand you have the set of the simulation set up on the other hand you have the basic results and of course you can see all other results in different formats if you like but this is what is very simple and they can see you have all the agents you can show them turn them off and also you can run your simulation based on default values that is what what is provided by the Department of Transport and also you can change the distances based on your expertise and what experts tells you to change so you can change and see the outputs and I’m adding to that some more functionalities but if you can please run and as I run this I explain some of these attributes I was not sure if I can use my laptop properly my laptop is really heavy in terms of the other that is in sometimes it crashes and I don’t want that to happen here so this is accrued I think this is crude all I can see it but this is a demo of that so you load your hazmat train this is basically when this this train enters the the city or near nearby and you load all your layers if you want you can turn them on and off this is ambulance for example and you run the default this is the default this is in higher speed so still is slow because train is this train is fan is moving really fast right now so as it gets near the city of course you see the impacted population based on those figures and on your right you have the risk levels by numbers and also the impacted population by by graph and you have all those tools these are GIS tools you can zoom in zoom out left right and up and down you can zoom into particular location of the the train as it moves you can zoom in and see the real path of the of the train I think I have some yes you zoom in I had the same issue with regard to the proper location of rail tracks sometimes but you can adjust that of course so here you see the color changes these are by the way the lowest level of census track these people are each indicating the the lowest census block sometimes they have an average of 500 people 300 you know varies from place to place and this also calculates schools this case I think the icons you see are schools that are impacted so if it is daytime it considers that impact if it is nighttime it also includes that that impact you can watch that but I explained the other options here what I’m trying to add to this is one area that is for for the derailment because we are interested for example emergency managers are interested in knowing what would happen if derailment actually happens in this

particular point so you make the derailment option it derails whether it’s night or day time when you push the derail button then you see the the simulation like none other stops but you will see the number of people in this case households that are within doubt threshold or impacted area and then I’m trying to simulate that evacuation from that particular impacted area so I don’t need to to evacuate the whole city or whole know that on on a specified area in the other hand I just focus on that impact zone and try to find out try to add those people as agents and work with them just to simplify and lower the so I think that’s where I stop the simulation for derailment assuming that a derailment happens here and then you can create a zoom out a bit and this is the the wind direction if you want to have particular wind wind direction and see who are the people within this window direction level I mean this is also similar kind of visual into this event but I’m working on it it’s not final but you zoom in these are the actual households that are living within that particular impact zone so you can actually simulate that the evacuation process can I stop this and move onto the next slide please yes that’s enough thank you yes future works what is missing in this is that the dynamics of day and night population is just one option you have day and night what I’m trying to do is to make it more dynamic so that at any given point of time and day using a function you have the actual number of people in in that particular day for a time for example if it is 8:00 in the morning you have the actual or estimated number of people based on down and I’m also as I said I’m adding the evacuation and shelter planning to this that’s the in progress also adding risk mitigation scenarios as I said if what would happen if you for example change the direction change the Train time and so on would would you be able to mitigate the risk based on these basic key policy options and also more intelligent into model for example if different for different types of hazmat you can have the characteristics of hazmat and build it into models so that when something happens you can have different dynamics and visualizations of the type of the impacts you might have and also training specifications and at this train as you see what just was for iconic tankers but you can actually model the whole train using it as an agent you have people can choose if this train has for example 10 cars 20 cars or whatever and which train actually which car is going to be Degrelle these are possibilities to be added to this and I stop here and leave some space for questions thank you very much thank you very much I just gonna remember when you were talking earlier did you talk about what is the proportion of derailments the result in rupturing the skin of the car resulting in actually release the rate is is is really low I have some some numbers here comparing for example two to two wrote there are more number of trained rear elements then for example wrote and also the pipeline’s but the portion is as it’s very low I mean the number is quite low I don’t remember on top of my hand but this is very low but overall the the thing is that when something like this happens the impact is is quite high that’s the difference but this is really loin have you considered putting the type of container car and some cars are better at containing spills and others based on the thickness of the hole as a factor in the likelihood of a spill when there is a derailment that’s a good question in fact that’s possible as you can see that training specification is kind of

important I think that’s what also the Government of Canada and so the u.s decided to change after this because of the the actual type of the the train and then the containers or tankers they were not really high quality so that in any case of derailment they leak so they have actually requests that stopping transportation of crude oil by these particular type a type of tankers that they were using so far unfortunately this lots of them are lots of rail companies are using this as snow but the regulation is going to force them to change quite fast but for now select 70% of crude oil is transported by this kind of regular tankers thank you