Episode 42. Drilling for Solutions: A gusher of new data is helping cut crash numbers in Texas’ Permian Basin.

Bernie Fette (host) (00:14):
Hello. This is Thinking Transportation. Conversations about how we get ourselves and the things we need from point A to point B — and all that can happen in between. I’m Bernie Fette with the Texas A&M Transportation Institute. A huge expanse of land in west Texas stretching into Southeastern New Mexico, is one of the biggest energy producing regions on this planet. The Permian Basin has long been known as a source of economic riches for the energy industry and for the communities and state whose fortunes are tied to it. But it’s also been known for something else, a remarkable frequency of vehicle crashes, many of them serious and deadly. That makes it a high-priority area for developing effective crash prevention strategies. Michael Martin, an assistant research scientist at TTI, is joining us for this episode to talk about efforts to curb the number of those incidents. Michael, thank you for taking time to join us today.

Michael Martin (guest) (01:20):
Thank you, Bernie. Glad to be here.

Bernie Fette (01:22):
You and your colleagues have been working on a very specific traffic safety problem — specific geographically because it’s confined to the Permian Basin in west Texas, southeast New Mexico, but we’ve had energy production fields, oil patches for decades. Why all the sudden interest and attention in those regions now?

Michael Martin (01:46):
Yeah, very good question. Most people would say, wow, you’re doing something really innovative in a very remote part of Texas, the Permian Basin in Midland/Odessa area of Texas. That yes is a very rural remote location, but it’s dealing with urban levels of traffic. And when you get with urban levels of traffic in a rural context that wasn’t built to handle that level, then you start to have issues, either access management issues or crash issues, safety issues, and, and such. Those were the leading points that got us here.

Bernie Fette (02:21):
When you started this work, what exactly was the intention? What were you setting out to do? And maybe you can talk a little about some of the specific problems you were looking at?

Michael Martin (02:31):
Sure. So if you can keep that picture in your mind of, you know, small roadways handling lots of heavy truck traffic and high volumes of traffic, there’s just kind of simple questions to ask and help TxDOT, the Texas Department of Transportation, answer. How do we improve accessibility while preserving mobility and making things safer? For instance, one road out there was called the most dangerous highway in Texas — U.S. 285.

Bernie Fette (02:59):
Was called that simply because of the raw number of crashes, injuries, fatalities, et cetera?

Michael Martin (03:04):
That’s right. Yep. And the first simple question was where are our driveways? Okay. And it was dealing with a high level of congestion and in particular locations. So along that roadway, TxDOT needed to know where folks were turning on and off of the highway, because where that happens, you have conflict points where crashes can occur, where slowdowns can occur, things start to stack up. And that was kind of the first question that we had to answer for them. TxDOT was having a lot of access management issues on U.S. 285, Texas 302, some of these major corridors that get you out into the oil patch. They’re carrying all the traffic from Midland/Odessa out to those locations with those issues. They’re also having traffic safety issues with crashes going up. That area of Texas has about 2 percent of the Texas population, but 8 percent of the highway fatalities. So, wow. It’s definitely overrepresented.

Bernie Fette (04:03):
So some of the specific problems that you were looking at, driveways was on the list. Can you give us a sense of, of what you kind of what you were looking at there?

Michael Martin (04:11):
Yeah, certainly. So, you know, we’re in far west Texas and the oil patch, there are access roads that intersect with the main highways. Those locations carry a lot of traffic and it’s typically large truck traffic, which moves slowly at first.

Bernie Fette (04:27):
And when you say access roads, you’re talking about access to the actual drilling sites, the well sites, et cetera?

Michael Martin (04:34):
That’s right. Yep. So just think of a white rock road coming and intersecting with the highway. They may look non-descript, but some of them, particularly in whatever state the wells are in, can be carrying thousands of trucks in a matter of a few days.

Bernie Fette (04:49):
Okay.

Michael Martin (04:50):
Some of the questions and issues we were trying to help TxDOT with were finding these driveway locations, figuring out how much traffic is on them and why are we having crash issues at particular locations? At the beginning, we used big data sources like Inrix. Then ultimately we turned to connected vehicle data that was much more frequent. Uh, every three seconds we get a dot on the map. So we could easily see where folks were turning, slowing down, hard braking, stuff like that. So it really helped to fill in the picture of what was going on.

Bernie Fette (05:25):
And where this started was again, with the driveways that you mentioned, the access points from the highways to the actual oil fields?

Michael Martin (05:33):
That’s right. Yeah. Some of the access roads were handling more traffic than some of the county roads, you know, formal roads out there, public roadways. And so not only finding the location. Coming up with some relative volume for these access roads was really useful. And you have to think, you know, expand that picture in your mind to a corridor that’s about 50 to 80 miles long. So going out there and collecting data in a traditional manner where you either hire students or hire a consultant to do it, would’ve taken a long time. Would’ve been expensive and possibly impossible. And you’re putting people in a very dangerous environment.

Bernie Fette (06:12):
You were looking at those driveways and what were some of the other major problems you were examining?

Michael Martin (06:19):
So being that it was deemed the most dangerous road in Texas, crashes were an issue. So what we tried to do is put those crashes on, on roadway segments so that we could figure out where are our problems occurring. You can’t just always look at frequency — you know, the number of crashes. You have to look at exposure there, people driving relative to the number of crashes.

Bernie Fette (06:40):
So how many of these safety challenges that you guys were working with are about the infrastructure that you’ve been talking about? You know, the roads, bridges, driveways. And how many of these challenges are about the people who drive on them? Most notably, the people who are working in those regions?

Michael Martin (06:59):
Yeah. That’s a perfect question for this project, because oftentimes in a situation like this, the traffic engineer may just look at how the roadway is built. Some of the geometry, physical aspects, the volumes, stuff like that. But in our case, we knew that the users played an important role, right? It’s not the roadways crashing into each other, if it’s people on the roadways crashing into each other. And so with the connected vehicle data and the in-vehicle monitoring system data — IVMS — we were able to pull out and observe driver behaviors in the environment in which they were crashing and having accessibility issues.

Bernie Fette (07:41):
You’ve got a road network that wasn’t built for such intense, heavy use. And you have drivers who are in many cases young. Perhaps very tired from very tiring work. And because again of their age, statistically predisposed to risk taking.

Michael Martin (08:01):
And also time is money. Yeah. It’s, they’re in a hurry. Right, right. Because if you don’t get to it, someone else will get to it.

Bernie Fette (08:07):
Right. Yeah. Drivers operating vehicles that are big and heavy under time pressures, et cetera, what could possibly go wrong? Is it fair to call this a perfect storm?

Michael Martin (08:18):
I think a lot of things aligned just right. Yeah. For these issues to be exacerbated by each of the points that you just made. And so you can’t just look at the physical infrastructure to try to solve some of these problems. You really do have to look at some of the behavior issues. Now, the data we use, it’s not tied to individuals. I wanna make that very clear that we’re not prying into anybody’s private life. We are simply looking at generalized behaviors that are occurring on a roadway segment or at a particular location. We don’t care if it was Joe Bob or Jane Doe. If we see high levels of hard braking at a particular location, along with a lot of rear end crashes, very likely to be strong relationship there, something is going wrong. And oftentimes when you see those sort of overlapping issues, it doesn’t tell you exactly the solution.

Michael Martin (09:14):
It just kind of highlights the problem, which 50, 80 mile long corridor narrows down, where you focus your efforts because we have a limited pot of resources to apply to these problems. So you have to be strategic. And that’s what we did. So we take crashes. We take the physical aspects of how the roadway was built. We take how many people are driving out there. And then we pull in behavior and overlap those layers, those on a map where you see problematic conditions overlapping, those are your target locations. That’s where you start to focus your efforts as opposed to, oh, we have a big problem of your, you know, the squeaky wheel gets the oil. Right. And so with this, it’s more of a data-driven approach and you can do a full-length corridor analysis in a matter of a month. And that timeframe is important because if you collected the data traditionally, by going out in the field, that would be far longer.

Bernie Fette (10:09):
So it sounds like what you’re doing, the layering approach that you’re taking is really giving a much clearer picture of the complexity of the problem. And to borrow your reference from just a moment ago, you’re giving some degree of intensity to just how squeaky the wheel is.

Michael Martin (10:30):
Yeah. That’s right. And whether or not the issue actually exists based on what we can see.

Bernie Fette (10:35):
Okay. Which again, gets you away from the anecdotal and into the more data-driven and data-precise understanding of the problem.

Michael Martin (10:44):
Yeah. We want this to be more proactive. Ultimately, a lot of what we’ve talked about is reactive, right? Issues occurred, now let’s react to them. Going forward we have the opportunity to use a lot of this stuff to be proactive, look at historical crash records. But then also look at either real-time or not that old behaviors to see, are we seeing precursors or leading indicators that these issues are going to either grow into a major issue? Or are we gonna, in fact, is this area that has no issues going to turn into a major crash problem based on some of the behaviors we’re seeing?

Bernie Fette (11:27):
Which would give you an opportunity to do what? Say place those access points in locations that made more sense so that they worked from the beginning instead of having to go back and fix the ones that don’t work?

Michael Martin (11:40):
Well that’s yeah. Good questions. The so what? Well, the, so what is, if you can start to better predict where these things are occurring, or going to occur Mm-hmm <affirmative>, you can go out there and treat it appropriately. Now that’s not necessarily my technical area. I’m more of a data nerd that is in the pit working on things to feed into these site types of analyses. But think about it in terms of signage, pavement conditions, geometry, I mean, how things are built. Do, does it still work with what’s going on out there on the road today? You know, are the users using it appropriately still? So there’s a lot of opportunity to try to make things safer. We just have to start to look in the right places.

Bernie Fette (12:30):
So you can see where you can make a difference.

Michael Martin (12:33):
Absolutely.

Bernie Fette (12:34):
You’ve outlined the challenges and the problems pretty clearly. So let’s turn our discussion to the solutions. You talked a little about how you use the data to figure out where to make access points to the oil fields make a little more sense. What are some of the other things that were done to address the crash problem?

Michael Martin (12:52):
For our latest study to use an example, it’s actually U.S. 385 south of Odessa between there and Fort Stockton, busy roadway. And it’s a four-lane divided, and they’re putting up a cable barrier median. This is far west Texas, the wild west. That median is pretty flat. A lot of folks drive pickup trucks, they’ll just make their own median crossing wherever. When you do that, you introduce a lot more conflict points. You introduce the opportunity for more crashes. We are already seeing some crash issues, uh, where we had more driveways, this exacerbated that problem. So with the cable barrier going up, TxDOT needed to know which crossings were used more than others, the formal ones and the informal ones that I described. The using of the connected vehicle data, we were able to actually pick out the U-turns and locate which ones were used more relative to the others, even if it was informal, it helped TxDOT to understand, okay, the public really thinks there should be a crossover here.

Michael Martin (13:58):
We can take that information and build it into our design process to determine whether or not we can make that crossover there, or how do we accommodate it some other way. That’s just one example of how we try to lessen the number of crashes and lessen the opportunity for crashes by reducing conflict points. Some of the other things were providing solutions, uh, back up north of this area in U.S. 285 to find some of these major intersections. So, a well access road, that’s carrying a lot of traffic. Remember a large truck stops, and then they have to start slowly by turning so that you now have traffic that’s supposed to be driving maximum 75, but we all have seen people driving higher speeds than that. And so they’re running up on a truck that’s just pulled out. It’s hard to judge speed, particularly in a wide open space like that. And so there, there were crash issues related to those locations. So what we did is we gave recommendations to put in right turn lanes, center, turn lanes, super twos. So that folks have the opportunity to get around these slower moving vehicles.

Bernie Fette (15:12):
When you say super two, Michael, can you help us understand what that is?

Michael Martin (15:17):
Yep. I’m not an engineer, so I may not use the right terms, but we’ve all probably traveled on one at one time. It’s basically where a two-lane highway turns into a three-lane highway and your lane may have an additional lane to get by, to pass. And then it’ll neck back down to two lanes. Right. And that can alternate from side to side.

Bernie Fette (15:34):
Yeah. And we might have seen those locations out in areas where we see the signs that say next passing lane, two or three miles from here.

Michael Martin (15:41):
Exactly, exactly.

Bernie Fette (15:43):
Right. Okay. Can you talk a little bit about the stakeholders who were involved? This is not just the energy industry having concerns with this. You talked a little about the Texas DOT. Can you explain to us a little bit about how the public sector and the private sector worked together on addressing these issues?

Michael Martin (16:02):
Yeah. We had a very positive experience related to that in this area. You know, you have very large companies operating in the oil patch and crashes take time and time is money to them. And everybody wants to make sure that everybody gets home at night. So the Permian Strategic Partnership, which is a grouping of these companies in this area that come together, they went and talked to TxDOT, how can we play a role in helping improve safety, improve accessibility and mobility out in the oil patch, cuz they knew they’re contributing to the issue, but they also wanted to contribute to the solution. So through that, we were able to negotiate and talk about some of the information that we needed. And so you heard me earlier talk about in-vehicle monitoring system data, IVMS data. That data came from six major oil and gas companies that said, “Hey, you need better information. We can give it to you.” And they worked with their IVMS partner, mixed telematics to scrub the data, anonymize it and hand it over to us so that we could fold that into our analyses and create that more complete picture.

Bernie Fette (17:21):
And once again, you were doing this in a way that did not involve any sort of personal identifiable information from the people who were driving those vehicles.

Michael Martin (17:30):
No, certainly not. And we do not want to touch any data that gives us personally identifiable information or PII. And so actually with the IVMS data, we didn’t have all the movements. It was just the driver events. So hard braking, hard acceleration for commercial trucks and passenger vehicles. And that may not be that impressive, but it truly was because the precedent that that sets in terms of, you know, major companies wanting to participate in a transportation problem by handing over data of how their employees operate on those roadways, that’s a big step forward in the right direction, in my opinion.

Bernie Fette (18:12):
That’s pretty big in the world of transparency,

Michael Martin (18:16):
I would say so for sure. Yeah.

Bernie Fette (18:18):
One of the things that you just mentioned when you were talking about the private sector partners in this picture, time is money and crashes cost money. The kind of safety problems that we’re talking about, both the number of crashes and the severity. They obviously have implications for human life. People get hurt, often badly, and sometimes people die in these crashes, but do the crashes also have a wider impact? And I think you were kind of touching on that just momentarily. I wonder if I could get you to address a little bit more about the implications of these crashes, the implications for the energy industry and for local economies.

Michael Martin (18:59):
Yeah. It’s a deep question, right? I’m not necessarily an expert for it, but I can give you my perspective.

Bernie Fette (19:04):
Sure.

Michael Martin (19:04):
From the company’s perspective, they don’t want any of their employees to get hurt, right? It’s terrible to deal with. But it also, if we get down to the bottom line, it costs a lot of money and it also drives up insurance rates. So some of these companies may be dealing with such great amount of issues that they may not actually be insurable any longer.

Bernie Fette (19:29):
Can’t stay in business that way.

Michael Martin (19:31):
Can’t stay in business that way. The greater impact to the economy is each time one of these crashes occurs in a network that already isn’t built out to carry the amount of traffic. It all breaks down. So a major crash will shut down a roadway for hours and there’s not good alternative routes to get around the crash. So…

Bernie Fette (19:54):
You’ve got ripple effects there too for the local economies about delaying shipments, delaying deliveries, et cetera. Yeah. Is fair to say.

Michael Martin (20:00):
I think that’s absolutely fair to say. Basically things need to occur when they’re scheduled to occur. Right? And so whenever you have these sort of breakdowns in the system, it can have a tremendous ripple effect for certain.

Bernie Fette (20:13):
You did this work in a very specific region that has its own unique characteristics, and no two traffic environments are exactly the same. But I’m wondering, can you take what you’ve learned in west Texas and apply it elsewhere?

Michael Martin (20:31):
Yeah, certainly. I think I said at some point, you know, this was really the first application for connected vehicle data. For some of these traffic transportation issues. John Speed at the TxDOT Odessa district was instrumental in supporting that research. You know, we were exploring new ways to view the problem. I keep using the analogy of painting a clear more detailed picture, but that’s exactly what we are doing with these innovative data sources. And so with that, you turn data into information and layering these pieces together, you get more information in a common location and you put subject matter experts, engineers, you put that in their hands and they can turn it into knowledge that they can act on. So all that can happen in a very short amount of time with these passively crowdsourced data sets.

Bernie Fette (21:28):
Yeah. You talked about how you can take data to the next step, information, and the next step knowledge or actionable intelligence. Can you give us an example or two of how that happens?

Michael Martin (21:41):
Yeah, my job as a data wonk is to take a lot of data points and turn that into something that somebody can actually understand and use. And that’s the information. I’ve worked with a lot of traffic engineers, transportation engineers, and so based on their training and professional experience, they can turn that into some sort of actionable intelligence. So an example here is connected vehicle data is millions or billions of points, points that you can plot on a map using latitude, longitude, coordinates, and each one of those points has some information behind it, like speed, heading, the year of the vehicle, something like that. Just some attributes to tell you a little bit about that point, taking that and turning it into some sort of meaningful result, like for instance, segment based rates. So not just looking at the number of things, but looking at it relative to let’s say hard braking, we don’t wanna just look at the number of hard braking events.

Michael Martin (22:45):
We wanna know the hard braking rate because it really depends on how many vehicles you have passing through there to make some sense and be able to compare locations. So that’s a meaningful result. Relative counts, knowing how much is occurring somewhere is also important. Particularly with traffic volumes. Turning percentages, U-turns, left turns. Those are really interesting events to pull out and give to somebody who’s designing a highway or dealing with a particular crash issue like rear-end crashes or left-turn crashes or anything like that. Speed profiles. Do you have a lot of high speed traffic moving in the same location as a lot of low-speed traffic when you have that separation or that sort of spread in speeds on the same segment, that’s opportunity for disaster because like I was giving you the example earlier, a truck pulls out and it’s moving very slowly getting up to speed, but somebody’s coming up at, you know, 75 miles an hour or greater.

Michael Martin (23:48):
You don’t want those two things happening at the same location. So you need to give people their own space. And then also looking at, you know, the connected vehicle data is a sample. It’s not all the vehicles out there. So we have to know what sort of sample rates we’re dealing with. And that can range between five to greater than 10 percent of the traffic. So that’s all meaningful information that paints that more complete picture of what’s going on out there on the roadways. So that busy engineers at the TxDOT district can take it and turn into some sort of actionable intelligence, a solution to resolve whatever issue that they may be dealing with.

Bernie Fette (24:30):
So what then is it that motivates you to show up to work every day?

Michael Martin (24:36):
Good problems like this. And what I mean by a good problem is working to find a solution to help save lives, time and money. I mean, it’s not just our motto. I mean, that’s a real thing that drives me. And a lot of folks that I work with. This is definitely a team effort and you know, working alongside other passionate people that have that common goal is what I think sets this work apart. I don’t wanna belittle anybody else’s work or beat down to it. It’s just, that’s what drives me. I like trying to solve problems.

Bernie Fette (25:14):
Michael Martin — assistant research scientist, and self-described data wonk at TTI. Thank you, Michael. You’re certainly helping to put a spotlight on this issue and we appreciate that.

Michael Martin (25:29):
Thanks Bernie. I appreciate it.

Bernie Fette (25:33):
The Permian Basin is big, in lots of ways. It’s one of the largest energy producing areas in the world, responsible for more than a third of all U.S. oil production. It’s big geographically too, covering more than 86,000 square miles. And it’s also been really big in terms of traffic safety challenges. But with careful research and targeted strategies, crash numbers are coming down. Thank you for listening. Next time, a special guest. Marc Williams executive director of the Texas Department of Transportation, will join us to talk about leading the TxDOT team, and he’ll share his thoughts on the future of transportation — in Texas and beyond. Please do join us for that conversation. And if you would please, give us a review, subscribe and share this episode. Thinking Transportation is a production of the Texas A&M Transportation Institute, a member of The Texas A&M University System. The show is edited and produced by Chris Pourteau. I’m your writer and host, Bernie Fette. Thanks again for listening. We’ll see you next time.