Railroad Grade Crossing Signal Systems - How They Work and Why They Fail

TASA ID: 950

On Thursday, December 13, at 2 p.m. ET, The TASA Group, Inc., in conjunction with railroad accident reconstructionist Robert W. Halstead, presented a free, one-hour, interactive webinar, Railroad Grade Crossing Signal Systems – How They Work and Why They Fail, for all legal professionals.

According to the Federal Railroad Administration, U.S. railroads currently have approximately 227,000 road crossings at grade, 140,000 of which are public roads and another 87,000 private. Approximately 53% of public crossings are equipped with flashers or flashers and gates. Collisions between trains and motorists at railroad/highway grade crossings are often catastrophic and occur somewhere in the U.S. an average of once every three hours, 365 days a year. Nearly 50 percent of crossing accidents occur at crossings that have flashers or flashers and gates. Although such automatic warning systems are extremely reliable, inadequate maintenance and/or testing can cause them to operate improperly, falsely indicating the approach of trains or providing short or no warning time to approaching motorists.

When handling a railroad accident case involving a crossing with automatic warning devices, it is imperative to have a basic understanding of the manner in which the warning devices are designed to function. Such an understanding can then measurably increase the quality of eyewitness interviews, help zero in on crucial information during discovery, and be invaluable to counsel throughout the case.

This webinar introduced basic information as to the workings of motion sensors and grade crossing predictors, then built on that information in a logical progression, discussed in sequence:

• Motion Sensors
  • How they work
• Grade Crossing Predictors
  • How they work
  • How Motion Sensors differ from Grade Crossing Predictors
• Grade crossing event recording capability
  • What to ask for in discovery
  • How to interpret crossing event recorder data downloads
• How Motion Sensors and Grade Crossing Predictors can fail
• Human factors considerations – how to question eyewitnesses and how to interpret eyewitness reports
• How Motion Sensors and Grade Crossing Predictors can fail and potentially cause or contribute to crossing accidents

About the Expert
Robert Halstead is an electrical engineer who served as Supervisor of Signal Construction and Maintenance on Consolidated Rail Corporation’s (Conrail’s) Albany Division, and is one of the nation’s leading experts on railway signal system design, maintenance, inspection and testing. He is an ACTAR-accredited accident reconstructionist, and has investigated and reconstructed more than 300 railroad-related accidents across the United States and Canada over the past 17 years. He is President of the National Association of Railroad Safety Consultants and Investigators, and guest-lectures to law enforcement and accident reconstruction organizations across the country.

Transcription:

Matt: Good afternoon. Welcome to today's webinar, Railroad Grade Crossing Signal Systems - How They Work, and Why They Fail. According to the Federal Railroad Administration, U.S. railroads currently have around 227,000 road crossings at grade, 140,000 of which are public roads, and another 87,000 private. Approximately 53% of public crossings are equipped with flashers or flashers and gates. Collisions between trains and motorists at railroads/highway grade crossings are often catastrophic and occur somewhere in the U.S. on an average of once every 3 hours, 365 days a year. Nearly 50% of crossing accidents occur at crossings that have flashers, or flashers and gates. Although such automatic warning systems are extremely reliable, inadequate maintenance and/or testing can cause them to operate improperly, falsely indicating the approach of trains, overriding short or no warning time to approaching motorists.

When handling a railroad accident case involving a crossing with automatic warning devices, it is imperative to have a basic understanding of the manner in which the warning devices are designed to function. Such an understanding can then measurably increase the quality of eye witness interviews, and help zero in on crucial information during discovery, and be invaluable and counsel throughout the case.

This webinar will introduce basic information as to the workings of motion sensors and grade crossing predictors, then build on information in a logical progression discussing in sequence, the following. Motion sensors, how they work. Grade crossing predictors, how they work. How motion sensors differ from grade crossing predictors. Grade crossing event recording capability. What to ask for in discovery. How to interpret crossing events. Recorder data downloads. How motion sensors and grade crossing predictors can fail. Human factors consideration. How to question eyewitnesses and how to interpret eyewitness reports. And finally, how motion sensors and grade crossing predictors can fail and potentially cause or contribute to crossing accidents.

The presenter for today's program is Robert Halstead. Mr. Halstead is an electrical engineer. He has served as a supervisor of signal construction and maintenance on Consolidated Rail Corporation's Albany Division and is one of the nation's leading experts on railroad signal system design, maintenance, inspection, and testing. He is an ACTAR, accredited accident reconstructionist, and has investigated and reconstructed more than 300 railroad-related accidents across the United States and Canada over the past 17 years. He is the president of the National Association of Railroad Safety Consultants and Investigators, and a guest lecturer to law enforcement and accident reconstruction organizations across the country. We will take two question and answer breaks during today's program. If you have a question, please use the chat or Q&A feature found on the right-hand side of the screen. We encourage all attendees to submit questions throughout the presentation.

Tomorrow morning, I'll send out an email with a link to the archived recording of this program. But we do ask that you take the time to fill out the survey that will appear on your screen after today's program is over. I now invite you to sit back, relax and enjoy. I'm going to turn the presentation over to our distinguished guest, Mr. Robert Halstead. Robert, the program's all yours.

Robert: I'd like to thank Matt and TASA for this opportunity to address you all. I'd also like to extend a warm welcome to all of our attendees. This is kind of an abbreviated presentation that I have delivered in the past at various points around the country, mainly to accident reconstructionists and law enforcement. I've spoken anywhere from an hour and a half to six hours, depending on the audience and the level of detail that they wanted to get into. But today, we're going to talk about 45 minutes or so on active warning devices and railroad grade process.

I think that Matt has pretty well covered most of the preliminaries, so I'm going to delve right into the meat of our presentation. Here's the types of cases that I typically handle. They involve not only grade crossings, but collisions between pedestrian issues, platform gaps, electrocutions, things of that nature. I'm going to dispense with the audience survey because we don't have any real good way to feedback here. But here's what we're going to do in terms of an agenda, which I think Matt has already spoken about.

So, there are different ways to classify railroad grade crossings, and we're gonna look at a couple of them right now. One way to classify them is by general types of usage. You have private grade crossings, those that are typically owned and maintained by private landowners, and you have public grade crossings, those which are installed and maintained by public entities of some kind. It's very rare, actually, to find active warning devices on private grade crossings. They do exist, but since they're privately funded, they tend to be fairly rare. Another way to classify railroad crossings would be as either a passive crossing, that is one that has crossbucks only. This particular picture has a stop sign. You know, it's not required that they do, some do, or you can classify it as an active crossing, that which has flashers, possibly also gates and cantilevers. And the photograph you see is actually a crossing that has all three at the same crossing.

So, every crossing that has active warning devices needs some sort of an enclosure to maintain the electronics and the electrical equipment that control the crossing. So, you're going to see either a crossing instrument case that you see in this photograph. The lower box behind it is a battery well. It contains backup battery power, or you will find a bungalow installed at the crossing. And actually, you're more likely now to run into a bungalow than you are a crossing case because a bungalow enables you to enclose the batteries right with the equipment that it backs up. And it also gives the maintainer a place to work on electronics outside of rain because, obviously, you don't want rain to get into your sensitive electrical equipment.

So, there are different types of train detection circuitry. We're only going to talk about two of them today. That would be motion sensors and grade crossing predictors. Now, I get the question a lot of times as to how a crossing actually detects the approach of a train. Does it sense the weight of the train, or does it measure the vibration on the rails as the train approaches, or does it use radar? The answer is, actually, all three of those have been tried at various points in the past. But far and away, the best approach that has been found to detect trains approaching crossings is to use electronics to inject current into the rails, effectively, and use the current passing through the rails to detect approaching trains. And that, in essence, is what a motion sensor and a grade crossing predictor actually do.

Now, this photograph shows a grade crossing predictor. This is inside the crossing bungalow that you saw in the previous slide. This unit is about the size of a microwave oven, perhaps two of them lined up next to each other, and it contains two duplicate sets of plug-in circuit cards. The left-most half is called the normal side of the unit. The right-most half is called the standby side. Now, these are identical, and normally only half of this unit is active at once, normally, the normal side. If the normal side fails for any reason due to an internal defect, the unit will automatically transition over to the standby side. So, it has some sort of limited fault compensation ability.

Let's see here. Okay. Now, we're injecting an electrical signal into the rails. We need some sort of way to connect from the motion sensor or the predictor to the rails. The way we do that is by burying underground cable from the bungalow up to the track. And we pop up through the ballast, and we connect to the side of the rail in that fashion. So if you go to a crossing that has active warning devices, you will see these track leads, we call them, at various points along the rail. Okay, and you can see it right there.

Now, the basic configuration of, say, a motion sensor or predictor, is as shown in this slide. The bottom box in the bottom of the screen represents the motion sensor. You can see we have a variety of plug-in cards. We have two pairs of connections to the unit. You have one pair called the transmitter, XMTR, and another pair called the receiver, the RCVR. And you can see that each of those are connected from the unit to the track.

The way the motion sensor works is it injects a constant current, AC, alternating current, signal into the rails through the transmitter lead. Now, those are connected 50 to 70 feet beyond the edge of the roadway. Okay? That signal travels down the track until it gets to what is called a termination shunt that is placed between the rails. It, in effect, sets up a tuned loop that extends from the termination shunt on one side of the crossing all the way to the termination shunt on the opposite side of the crossing. And depending on the speed of the track, you may have well over a mile between those two shunts. So, the predictor is setting up a tuned loop that is constantly circulating, looking for trains that extends from the left-most shunt in this picture, to the right-most shunt.

The receiver leads are plugged 50 to 70 feet on the opposite side of the highway, and those are essentially a voltmeter. They just sit there, and they passively measure the voltage that is generated across the rails as a result of the electrical signal that's been in. Okay? And the distance in between where the transmitter leads are connected to the rail and where [inaudible 00:12:40] connected is called the island circuit, and we'll discuss the island circuit separately a little bit later on.

All right. Let me go to the next one here. Okay, this is a view of an actual termination shunt. It looks like a coffee can, maybe two of them stacked on top of each other. Two wires come out the top. One wire plugs to one rail. The other wire connects to the other rail, okay? And what that does is it serves to terminate that circuit. It looks like an electrical short circuit for the frequency that this crossing is putting out. But it's invisible to all other frequencies in the track.

Okay, so if we look at a voltage versus distance graph up above, we can see that with no trains between the termination shunts or the crossing, the voltage is at a constant high level. That's the voltage as measured by the receiver pair. Now, as the train moving from left to right crosses the termination shunt and enters the eastward approach circuit, you can see that the voltage, as measured, is beginning to decline. That is because the train is electrically shorting so that the current being injected into the rail is not making it all the way down to the termination shunt and coming back. But instead is jumping up the wheel, across the axle, and down the other wheel of the leading wheelset that's approaching the crossing. And that causes the voltage to decline at, hopefully, a constant rate, a straight smooth line as you see there.

Now, as the train continues toward the crossing and actually enters the island circuit, you can see that the voltage measured by the receiver leads declines to zero. That is because you now have a dead electrical short between the transmitter and the receiver leads. Okay? A transmitter can pass cannot transmit through a dead short. So, the receiver sees nothing. Okay? That's how the motion sensor, or the predictor, determines that the train is physically on top of the highway, that it has occupied the island circuit. Okay, if we continue to the right, we can see that as long as any portion of the train is in the island circuit, i.e., across the highway, the voltage as measured by the receiver remains zero, okay, because you still have at least one or more electrical shorts between the transmitter and receiver.

Now, as the train continues to trail away from the crossing and the rear of the train clears the island circuit, now, the receiver is able to see signal transmitted by the transmitter once again. So, the voltage begins to increase. And the voltage will continue to increase until the train exits the most distant shunt there, at which point the voltage between the rails returns to the same high level that it was before the train arrived. Okay? So, at this point, the crossing has now entered its quiescent state. In other words, the state that it was at before this train ever showed up on the scene. Okay?

I know we've kind of gone through that a little fast and I apologize for that. Certainly, if you have any questions or would like any clarification on that, please don't hesitate to do give me some Q&A on that, and I'll be happy to address that.

Now, how far do we place the crossing termination shunt from the crossing? Well, that depends on a couple of factors. First, it depends on the maximum authorized speed on that section of track. In other words, the fastest train that you should ever expect to be approaching that crossing. Okay? Once we know that, we also need to know the warning time that we would like to give for that train, be it 20 seconds, 30 seconds, 35. And we can take the train speed in feet per second, multiply it by the desired warning time, and that tells us how far from the crossing we need to locate the termination shunt.

For instance, if we want to provide 30 seconds of warning time to 79 mile-per-hour trains, then we need to locate the termination shunt at least 3,476 feet prior to the crossing. Now, actually, a predictor requires four additional seconds for what we call acquisition time. It needs time to acquire the inbound train and calculate its speed. But we'll discuss that a little bit in some further slides.

Now, if you have a motion sensor here, as soon as the train crosses the termination shunt, that's it. The motion sensor detects and it activates the warning devices. So, if, for instance, you have a train coming along at a speed that is significantly below the maximum design speed, then by default, you're going to get a higher warning time. For instance, if you have a 40 mile-per-hour freight train approaching that very same crossing, a motion sensor is going to give you 60 seconds of warning time instead of 30 because it takes the train twice as long to cover the distance between the termination shunt and the crossing.

Now, that situation gets even worse. If you have a work train, say a 20 mile-per-hour work train entering a motion sensor circuit it's going to take a full two minutes to travel from the shunt to the same crossing. Now, there are a number of federal studies that have been done that have linked warning time to driver compliance. And, in essence, they have found that when the warning time begins to exceed roughly 45 to 50 seconds, the rate of driver noncompliance tends to rise markedly. Okay? So, motion sensors are good to a point, but they do have this problem.

Let's forge ahead. Now, if a train enters a motion sensor or a predictor approach, and then stops in the approach before it gets to the island circuit, then the declining voltage curve suddenly levels off. And the motion sensor detects that leveling off and it says, "Oh, it must be the train stopped." So, the motion sensor monitors that train. And after that train has stopped for at least 10 seconds, it will go ahead and recover the gates. In other words, gates come up, the flashers shut off. But it continues to monitor that train for signs of renewed movement. And if it senses that the train has renewed movement toward the crossing, in other words, the voltage curve goes from leveled off and begins to once again decline, then, it will go ahead and reactivate the warning devices.

Okay? But if the train resumes movement away from the crossing, then, the voltage curve goes from leveled off and begins to increase. And when it increases, the motion sensor predictor knows that the train is moving away from the crossing, and therefore, it will not reactivate the warning devices. So in essence. In most cases, a motion sensor predictor has no idea what direction the train is approaching the crossing from. It only knows if the train is approaching the crossing, or if it's trailing away from the crossing by whether or not the voltage curve is going up or is going down.

Now, at this point, again, I know I've gone through things pretty fast. But if anybody has any instant Q&A, I'd be happy to address that, or, you know I can hold it till the end, whichever you prefer.

Matt: Oh, Robert, we do have a couple of questions here that have come into the queue and I would encourage all in attendance to submit their questions either by using the chat feature or the Q&A feature. Robert, during the beginning of the presentation, you talked about the normal and the backup within the box. If it were to fail, does the backup kick in instantaneously? And are there any other redundancy tools built into the system?

Robert: Yeah, that's a good question. The very most basic units do not have a redundant feature. They're equal to only one-half of the unit that you saw on that photograph. But if an error pops up that the normal side can't deal with, it will attempt to transfer control to the standby. And that process will take about 30 to 45 seconds, on average, during which the warning devices will be activated. Flashers on, gates down. Now, if the standby side comes up, but sees the same problem that caused the normal side to go down, then the standby will go down as well. In other words, both the normal and the standby, for instance, are both connected to the track, the same track. So, if what caused the normal side to go down is a failure somewhere in the track, then when a standby comes up, it's going to see the same error, and it's going to go down and try to transfer control back to the normal side. And then the normal will come up, see the same problem, try to go standby. And basically, the crossing will continue to go back and forth until someone gets there to look at the problem.

Matt: Okay, great. We have a question here from James who asks, "What is the minimum required warning time? And does that mean when the lights and bells start, or does it mean when the gates start to move?"

Robert: Yep, another good question. And, in fact, I have several slides further on that address that very issue. So, let me hold that and we'll get back to that question.

Matt: Okay, great. Where I live there's a railroad crossing right by the center of town, and my wife hops on the train every day. And I've noticed that people try to outrun the trains or hop in front of them as the gates are down to catch the train. You know, what's the perception that the... I mean, how fast is the train usually moving at a crossing, and how much time does do people have to truly get across the tracks? I think that's a huge liability for the train companies.

Robert: Well, I can address the question on several levels. First of all, you know, the deal with eyewitnesses, will be able to tell you that oftentimes eyewitness estimates of time, speed, and distance can be off, sometimes widely off the mark. We see that a lot in the railroad because you have a large, heavy, moving object. It's, kind of, like an airplane. You see it coming in for a landing and it tends to float in the air as it comes through, even though it's moving 200 plus miles per hour. The same thing with a railroad. So, oftentimes, people tend to radically misjudge the amount of time that they have to complete a crossing transit maneuver, be it on foot, or be it in a vehicle. Warning device credibility. Well, actually, let me hold off the second part of that question until we get closer to the end because I have several slides that deal with that as well.

Matt: Okay, excellent. Robert we've had some people who have commented that you're breaking up a little bit or you're a little...you're not coming through as clear as possible. Could you move closer to your speaker, possibly, in your speakerphone?

Robert: Okay. Actually, is this better?

Matt: That's a lot better. Yes, thank you.

Robert: Okay. What I've done is I've gone from my headset to just talking in the basic handset here.

Matt: Yeah, that's a lot better. Why don't we continue on with the presentation of content?

Robert: Okay, great. I apologize for any audio problems. Up to this point. I'll try and do better here. Let me see. All right. Let's continue on here. Okay. Now, the first part of our presentation we discussed mainly how grade crossing devices operate. Now, we're going to kind of extend that knowledge and explore some possible ways that they can fail to operate safely, or operate as intended. And there are two principal ways that grade crossing warning devices can fail. The first type of failure is a false activation. That is when the warning devices are activated, and there is no train approaching the crossing.

Now, the second type of failure, which is much more serious in a direct sense, is the failure to activate. That is when you have an approaching train with either short warning or no warning time. And there is a technical term in the industry that we used to refer to that, and that is a whoo whoo with no ding ding, not a good thing.

So, let's discuss each of these. Now, here's a slide that addresses one of the previous questions. What is the required minimum warning time? According to 49 CFR Part 234, a crossing must give at least 20 seconds of warning time. And that is measured from the instant that the first light begins to flash until the instant at which the lead locomotive physically occupies the highway. It must be no less than 20 seconds. A second federal regulation, also contained in Part 234, is that if the crossing is equipped with gates, the gates must be fully horizontal for a minimum of five seconds before the train physically occupies the highway. Now, given the 22nd requirement, there's no railroad that designs their crossing to give exactly 20 seconds of warning time. You always design a crossing for at least 25 seconds of warning time. In fact, most, a lot of gate crossings are designed for 30 seconds of warning time just to make sure that you never, ever fall below the 20-second requirement.

Okay, here's kind of an ancillary slide. If you have [inaudible 00:29:25] malfunctions crossing, here is the maximum speed that a train may traverse that crossing. For instance, if you have a reported activation failure, and you have no flagger or law enforcement presence, trains must stop and manually flag their way over the crossing. I recently handled a case, in fact, where the crew was required to do that, but they got the crossing mixed up. They thought it was actually... You know, the crossing that they had to stop for was several crossings beyond the one that they actually did. So, instead of stop, they entered the crossing at 38 miles per hour and collided with a vehicle resulting in, unfortunately, two fatalities. So, when these requirements are not adhered to, you can have some serious consequences.

Now, just to give you a brief flavor of some of the data that you might be able to get out of motion sensors or grade crossing predictors, this slide shows the most basic type of data download that you can get. This came from a Safetran 3000 system. It's a grade crossing predictor. It's pretty common around the country. This is an example of a download. For instance, you have different columns that show date and time. Input channels, we'll ignore for a moment. And then, you have four columns that show speed. Actually, three that show speed. The first one is called WT. That is warning time. That's the warning time that the crossing recorded, that it gave.

Let me take a moment and make a very important distinction here. This basic download, as you see it here, and the warning times that are presented there reflect only the amount of time that elapsed from when the predictor instructed the external circuitry at the crossing to activate the warning devices and the instant at which the train entered the island. So, in other words, this records only the time that it gave the command. It does not tell you whether or not the external circuitry that it commanded actually carried out the function. Okay, a very important distinction.

Now, interesting if you look down the warning time column, you'll see a 10. Okay? We have what appears to be a very screwed-up train. On July 17 at 21:00 hours, it got only 10 seconds of warning time, which is significantly below the federally-required minimum.

Then, if we continue looking at columns to the right, we see DET, AVG, and ISL. DET is the detect speed. That's the speed that it calculated the train was moving as it crossed the termination shunt headed to the cross. In other words, when it first acquired the inbound train. Then, the ISL is the island speed. That's the speed at which the train was traveling as it entered the island circuit, which is effectively the highway. And then the AVG is the average speed over the entire approach distance.

Now, I can caution you, as you're looking at these speeds, these speeds are not accurate. They're intended for maintenance purposes only. If you want truly accurate speeds, or well, probably the most accurate you could get, you would have to look at the locomotive event recorder download. If the wheel size was measured correctly, those speeds will typically be much more accurate than what you'll get out of a crossing event recorder.

The principle of use, LD speeds as you see here, is to determine whether the train maintained a pretty constant speed as it traversed the approach, or if the train accelerated markedly, or if it decelerated markedly. For instance, you can look at that record that has the 10-second warning time. And you can look at to detect speed was 55 miles per hour, but yet the island speed was zero. So what that's telling us is the train entered the approach at 55 but slowed all the way to a stop by the time it got to the crossing. And I think that's very suspicious. I don't recall the exact case that this came from but there aren't a lot of trains that can perform in that fashion. So, that would tend to indicate that there may have been an error within the unit. And, in fact, you can see an error code of 9011 was recorded.

So, moving... I'm sorry. Let me go back to that and mention one more thing. The input channels that we skipped over, you can see they're numbered 1 through 8 and 9 through 16. The GCP 3000 system, as is true with most motion sensors or predictors, has additional event recording capability for events that occur outside of itself. For instance, I told you about external circuitry. When the motion sensor detects a train, it notifies the external circuitry that it's time to activate the flashers and gates. What these input channels are intended to do when they're hooked up is record whether or not that external circuitry actually obeyed the commands that it was given. And this particular download, you can see that all those input channels are blank. So, this crossing is not recording anything that occurs outside of the predictor unit.

Okay. I've seen that, as time goes on, we tend to actually get more and more event recording capability designed into crossings. For instance, many of the current crossings I can tell not only that the flashers were physically on, but I can determine the rate at which they flashed. I have a much higher level of detail available to me. Unfortunately, that generates a much more complex data download, and one that you really need signal expertise in order to be able to fully interpret. I just wanted to show you a very basic issue here. Let's see. Then, I have a quick list of crossing event recorder data to request the event log, which is what you just saw on the previous slide. You want to ask for that for every motion sensor or predictor of the fault or error log. You saw that one error message that was logged. That could be important. And the application programming history. That tells you how the unit was computer programmed because, basically, it is a computer. And at various times, I have actually found these units misprogrammed in the field and misprogrammed in a way that causes them to function in a manner less safe than designed. So, it's always important to check the programming.

Now, a motion sensor, as we talked about, it injects an AC signal of a specific frequency into the rail. And if you have crossings that are close enough together that causes their approach circuits to overlap, that's not a problem because you just put the crossings on different frequencies. There's, I think, 9 or 11, different frequencies that each manufacturer has. They commonly call them channels. As long as each overlap crossing is on a different channel, you don't have any overlap issue. Again, it's easy to overlap approach circuits. And a motion sensor as we indicated, when a train stops for more than 10 seconds, it causes the crossing to recover. Not the case in the older DC-type circuitry that it replaced. In that type of circuitry, as long as the train is on the approach circuit, flashers are on, gates are down. Okay. We're not going to talk about lockout failures. That's a more detailed subject.

Let's continue here. Let's see. Okay, disadvantages. Now, this is the principal difference between a motion sensor and a grade crossing predictor. With a motion sensor, anytime a train is within any portion of the approach circuit, the crossing warning devices are activated. So, you can have that problem where you have the really slow train with a really long track circuit can cause the gates to be down for an inordinate period of time. The way a predictor works is it detects trains in exactly the same manner as does a motion sensor. But it takes that information one step further. It analyzes the slope, or the rate of decline, of that voltage curve and calculates from that the speed of the approaching train. It then computer calculates a delay time. And it will delay the activation of the warning devices as necessary to achieve a constant warning time at the crossing.

For instance, if your approach circuit is set up from 79 miles-per-hour, and you have a 40 mile-per-hour train enter that same circuit, the predictor will allow that train to come roughly halfway into the approach circuit before it activates the warning devices. So, now, that can throw some eyewitnesses off because if they're used to seeing warning devices activate with a train a certain distance from a crossing, and you have a lot slower train approach the crossing, it can really freak people out because they see a train much closer to the crossing than they're used to without warning device activation.

Now, one other point. You program into the predictor your desired warning time. And the predictor will do its best to try and meet that designed warning time, but it won't be perfect. You'll typically see anywhere from two to five seconds, plus or minus on either side of your designed warning time. So, for instance, if you programmed it for 30 seconds, you'll see some trains with 32 seconds, 33. You might see a 37 or two. You might see a 27. So, it's gonna try and come as close to that as it can. But it may well not hit that. Let's see here.

Okay, a quick discussion of the term fail-safe. Whenever you talk, for instance...If you're on the plaintiff side, and whenever you talk to a rep talking about the operation of the warning devices, they will assure you that the warning devices could not possibly have failed because they are, "fail-safe." And the term, fail-safe, is a design philosophy. It is not a performance guarantee. What that term means is that crossing circuitry must be designed on what's called the closed principle, or closed-circuit principle, such that a failure in almost any portion of the circuit must cause the warning devices to assume their safest condition. And that is flashers on and gates down. For instance, if you have a rail break, that will break that loop track circuit, and it will cause the warning devices to activate and the gates to come down. And they'll stay down until the defect is repaired.

So, now, also, the term fail-safe, when applied to a crossing is flashers on and gates down. Now, that does not, however, indicate that the crossing itself is "safe" because you run into issues of [inaudible 00:43:11] or a motorist warning device credibility. If they see warning devices activated for long periods of time with no obvious train approaching, it tends to reduce the credibility of the warning device in the eyes of the motorist. Okay, and that is not a safe thing. So, I want to make the point here that "fail-safe" is not necessarily safe.

Okay, proceeding on, let's look at some possible causes that can cause a crossing to give either short or no warning time. The first one is a piece of band iron. Lumber shipments on the railroad are typically secured by a steel strap that encompasses the lumber. And sometimes as the load shifts, those bands tend to break. And they may come off the train entirely and be strewn along the right-of-way. It's not at all uncommon for pieces of that to be grabbed by approaching trains and inserted into the track in such a way that it bridges the [inaudible 00:44:24]. And when, you know, steel or anything electrically conductive bridges the two tracks, it forms a short circuit. And the crossing immediately interprets that as a stopped train at that point.

So, because the voltage measured on the track immediately declines and then levels off. So, the crossing monitors that, and if that's present for 10 or more seconds, it goes ahead and recovers the crossing. But the crossing will not be able to detect any trains beyond that point. So, if you have a piece of band iron that bridges the two tracks 100 feet from the crossing, the motion sensor or predictor is unable to look beyond that for approaching trains. So, if a train does approach. It's not going to pick it up until the train crosses that point, which may well be much too late to activate the warning devices. That's something to look for.

The other thing is rail slivers. You'll see these are particularly on track circuits that are on a curve. Heavy trains will tend to shave off a thin layer of rail, typically on the high rail on a curve. And I've seen, you know, rail slivers that exceed 20 feet in length. And they can get worked up just right, and get between the rails, and cause the exact same problem that you can have with band iron, rendering the crossing unable to detect anything approaching beyond that point.

A shorted front rod in the switch I'm gonna show you right here. Okay. Here's a switch in the track that you're looking at. You can see about the third tie in, you have a metal rod that goes from one track or one rail to the opposite rail. And you can see a green thing in the middle of that. That's an insulated front rod. All that is is a metal rod. And although it's not insulated, it will electrically bridge one rail to the other and create a short circuit, again, rendering the crossing unable to detect anything beyond it.

Let me scroll up here, and we'll get to the next slide. Well, let's see. Okay, and there's an example of what they actually look like in the field. Okay? It's a solid bar that's been cut in half and spliced with an insulated splice that maintains the mechanical integrity of the rod but keeps each end of it electrically isolated from the other. And if that insulation... And you can see that some of that insulation is not the best looking there. If that deteriorates to a point where that shorts out, then it produces a situation similar to the band iron that we saw previously. Okay? You can also have bad insulated joints that can cause short or no warning time. This is the same switch I'm pointing to. Right here, this is called the curved closure rail of the switch. You can see it extends from the left rail to the right-most straight rail. And if that was not electrically severed, then that would create a dead short between these two rails. Okay? So, if the insulation in this insulated joint becomes bad, you can have, again, cause the inability to look beyond it.

A rusty rail, okay, this is a big one, especially on branch lines. If you don't have enough train traffic or heavy enough trains to keep the rail nice and shiny, what you can have is you can have the build-up of rust on the top of the rail. And rust is a very good insulator. It interferes with the electrical continuity between the wheels of the train and the track, which can cause a failure of the train to achieve shunt. That's what we call it. In other words, to electrically short one rail. And if the rail doesn't short... Sorry, if the train does not short one rail to the other, then it becomes invisible to the crossing. The crossing has no idea that it's approaching. We also see this problem on passenger trains because they're light. You don't have a lot of axles there. Passenger trains oftentimes use disc brakes. They don't use the normal method of sliding a brake shoe along the bearing surface of the wheel, so they don't keep the wheel surface nice and shiny.

Also, I have another slide coming up here that will give you another example. Let's see here. Let's try to go to the next slide. Let's see here. Okay, let me go to the next one. Okay, this is an excerpt of an NTSB report on the Bourbonaisse accident, which occurred several years ago. Amtrak hit a loaded steel truck at a crossing. I believe 14 fatalities. NTSB is noting that they personally observed two activation failures since the collision. And 6 seconds of warning time is what they observed on one train as compared to the required minimum of 20. The second one was due to leaking product out of a car that was building up on the track. And the train rode up on that pile of material causing it to break electrical continuity and causing it to effectively disappear, as far as the crossing was concerned. Okay, something we always look for.

Maintainer's jumper. Sometimes when the signal maintainer is out at a crossing, they need to perform a maintenance or a task that requires the crossing to be activated. And particularly, if you have a very busy highway, you want to try to minimize the amount of time that the crossing is activated with no train approaching. So, they'll take a short piece of wire, and they'll connect it from the positive side of the backup battery to a certain point on the control circuitry that will bypass the approach circuit. And while that jumper is in place, the crossing will not be able to detect any approaching train from any direction. And that situation is okay, as long as the maintainer has permission to do that, has coordinated with the train dispatcher, and is physically on-site to manage that situation.

But every now and again, a maintainer will leave the crossing and forget to remove the jumper. And when that occurs, the crossing will be unable to detect the next train that approaches that crossing. So, if you get eyewitness reports that say that there was no activation of the warning devices whatsoever on the approach of this train, not even when the train was in the island circuit, then this is what I suspect is somebody had a jumper on it.

Okay, the voltage output through the transmitter leads. It could be that the maintainer adjusted that but did not do a proper checkout afterward to make sure that he wasn't overdriving the circuit. Because if you overdrive the circuit, it's possible that you could be pumping so much current into the rail that you can bleed by any shunts that are present on the circuit. And I actually had a recent case where we had exactly that. Let's see. Now, here's an example of manufactured warnings. Okay, check-out procedures must be adhered to. Failure to do so may result in reduced warning times. Okay, sometimes no warning time.

Just a couple of excerpts here from, let me see, okay, proper adjustments, again. This is what the manufacturers put out in their manuals to railroad signal personnel. Okay, here's more example. RF level, reduced warning time. So, what these are, are examples of the manufacturers themselves telling the railroads that if you go in and change adjustments here without proper checkout, you may well create an unsafe condition at the crossing. Okay. Failure to comply with manufacturer's directives.

Now, manufacturers periodically put out what are, in essence, recall notices like car manufacturers do saying that, you know, "We've observed a certain problem, and it's critical. And here's what you need to do to fix it, and here's the level of urgency that you need to fix it with." Okay. Here's an engineering service bulletin as an example. Let's see. Urgency on this one is as soon as possible. Okay. And now, you can see this failure can result in complete activation failure of the crossing. And what I've found in many, many cases over the years is that many manufacturers actually have no internal means of tracking their response to these recall notices. And they have no way of determining whether a particular crossing has ever been fixed, according to a given recall notice. So, as you can imagine, this would be a problem, on occasion.

The one thing that I do, for instance, is I examine the motion sensor at the crossing very carefully, and I record the serial number of each of the individual circuit cards. And I run those against a recall list to see if any of them have ever been the subject of a recall, and if so, whether or not the required change has ever been made.

Another cause of short warning time is the fact that the engineer accelerated on the approach to the crossing. If you have a good crossing predictor, it calculates the speed of the approaching train during the first four seconds that the train is on the approach. It basically samples three and a half seconds, generating eight samples, which it then averages, and uses that resulting speed to calculate that much delay it should observe before activating the warning devices. So, if you have a train come in at a given speed and activate, you know, the warning devices, and then significantly accelerate toward the crossing, the train may actually arrive at the crossing, before the gates are fully down before the predictor calculated that the train would get there. And almost all railroad operating rule books require trains to maintain a constant operating speed with [inaudible 00:56:41] approach to deal exactly with that problem.

Let's see here. Okay, now, far and away, the most common cause of active warning device failure at grade crossings is a loss of commercial power, with a subsequent exhaustion of the battery standby capability. Crossings are required by federal law to have battery backup capability. The CFR, federal regulation, does not specify, however, how long that backup has to be able to hold the crossing. So, what we've seen, especially at flasher-only crossings, is commercial power was lost at some time in the past. The batteries were exhausted. And therefore, the next train that approached, which was the one that was involved in the accident, got either very dim flashers or no flashers at all because there simply wasn't any power present at the location to flash them.

Now, if you have a crossing that has gates, gates are designed in such a way that it requires power to hold them vertical. So, if both commercial power and standby power is lost, the gates will be released to slowly drift down to the horizontal position where they will remain. So, you'll have a dark crossing with the gates horizontal, and that's it. That's, you know, technically the safest possible way that you could fail the warning devices at that crossing. And it's trusted that, you know, somebody's going to see that, and they're going to report it to the railroad, and get that taken care of.

So, a few things to obtain in every investigation, of course, is prior weather reports. You want to talk to the power company, see if they've had any power outages or ground fault interruptions that might have affected the power supply to your crossing. Let's see here. Okay. Human factors. I'm just gonna deal with this very briefly. I could probably spend an hour on each one of these. But drivers expect crossings to look and function in essentially the same manner every time. And that can be the same crossing every time or a different crossing. They should all function in, essentially, the same manner. And, you know, when you violate those expectancies, oftentimes, drivers react unpredictably. And it's that unpredictability that oftentimes either causes or contributes to accidents.

Warning device credibility, what we talked about. When warning devices are activated, and drivers expect there to be a train approaching when there isn't, you know, they get miffed. And they assume the next time they see activated warning devices that a train probably isn't coming there, although it may be. Another thing I can inject, false activations often occur during periods of rain. So, if you talk to your eyewitnesses and they say that, "Yeah, every time it starts to rain, the gates come down," that tells you that you have inadequate track maintenance because what happens is you have dust and dirt that is in the ballast, the rock, that's underneath the track. And when you mix that with water, it becomes an electrically-conductive slurry, which tends to simulate the standing track, in effect, and causes the warning devices to activate. But as it stops raining and the sun comes out and dries the track circuit out, oftentimes, the crossing will go right back to work.

Perception reaction time. I typically use one and a half to two and a half seconds for perception reaction times. One and a half is probably closer to, you know, the 80th percentile, and two and a half seconds is what federal and state highways are designed for. Usually, you know, I don't have a problem if I assume something in that area. But you can see a driver with a situation that they don't expect. Oftentimes, their perception reaction time will increase. And the more unexpected it is, oftentimes the more it will increase. Let's see here.

Okay, here's an example of an integrated timeline. This is what I did for Fox River Grove. That's an accident where commuter trains [inaudible 01:01:48] with fatalities, unfortunately. But as you go from left to right, you're approaching the crossing. Starting 40 seconds from impact, the one line that is predominantly green shows you the actions of the engineer over that time span. You can see that he did not apply brakes at all until about 10 and a half seconds. I'm sorry, eight and a half seconds before impact. And seven seconds before impact, he went into full emergency. But at that point, it was far too late. Up above that, you show the operation of the warning devices that are synchronized to that. About 31 and a half seconds before impact, the predictor first picked up the approaching train, having what we call acquisition time.

Then at the end of that time, that predictor calculated that it needed to delay three seconds. So, you have a delay time of three seconds in order to get the desired warning time. Then you had one second of relay response time. Then you had gate delay time, three seconds. This addresses a previous question. I think it was to the effect of once the flashers begin to flash, do the gates immediately begin to come down? The answer to that is no. It's federally required that there is a three-second delay between those two events. I personally have seen up to six seconds in the field. It has to be at least three. Okay? Then the yellow trace on that line is the gates are actually descending until you get to a point five and a half seconds before impact, at which the gates are fully horizontal. And down below, you can see the train's distance to impact at each of those one-second intervals, as well as its speed.

Now, this particular accident, I think, was primarily caused by a traffic signal controller that was controlling an intersection that was within 30 feet of this crossing. So, the two are interconnected. And I didn't have the data in time from that controller to put on this diagram. But it'd be pretty instructive to also show what the traffic signal was doing throughout this entire accident sequence. That can be done as well.

Okay, here's a handful of information sources that are commonly used for investigation and reconstruction in the railroad environment. I encourage you to, you know, look through them. And you'll find some great research reports on human factors and a wide variety of other topics. Okay, it's now... I'm sorry I ran just a little bit over here but I'd be happy to address any questions that you have. Go ahead, Matt.

Matt: Okay, great. I don't see any questions in the queue. So, what I'd like to do is just wrap things up here very briefly. Robert, thank you for the time and the effort that you put into this presentation. I think you did a great job and presented the attendees with a lot of very useful information that they can take with them and use on their cases. If you'd like to speak to Robert about a specific case or project that you're working on, you can contact us here TASA. Our telephone number is 800-523-2319. And tomorrow morning, I'll send out a link to the archived recording of this program. We post all of the archived recordings of all of our previous programs in TASA Knowledge Center. Go to TASAnet.com and click on the Knowledge Center tabs at the top of the page.

If you have any follow-up questions or comments, please feel free to email me at exerts@TASAnet.com. You'll be receiving an email at 3:30 this afternoon from me thanking you for attending. You can simply reply to that, or like I said, at any time contact me at experts@TASAnet.com. Otherwise, I thank you for your time and I wish you a safe and happy holiday season. And we look forward to seeing you back here in 2013. Thank you so much for your time. We look forward to seeing you at future TASA events.