View the March 17, 2021 seminar recording
Jennifer Symoun
Good afternoon or good morning. Welcome to the Talking Freight Seminar Series. My name is Jennifer Symoun and I will moderate today's seminar. Today's topic is Trending Research in Freight Transportation Planning and Logistics.
Before I go any further, I do want to remind you to call into the teleconference for the best audio quality. If you are listening to the audio over the computer and experience any issues, I am unable to fix them as audio quality will vary based on your network connection, computer, speakers, and other factors. Please also keep in mind if you are calling into the teleconference for the audio, you will need to mute your computer speakers or else you will be hearing your audio over the computer as well.
Today's seminar will last 90 minutes, with 60 minutes allocated for the speakers, and the final 30 minutes for audience Question and Answer. If during the presentations you think of a question, you can type it into the chat area. Please make sure you send your question to "Everyone" and indicate which presenter your question is for. Presenters will be unable to answer your questions during their presentations, but I will start off the question and answer session with the questions typed into the chat box. We will also take questions over the phone if time allows and I will provide instructions on how to do so once we get to that point.
The PowerPoint presentations used during the seminar are available for download from the file download box in the lower right corner of your screen. The presentations will also be available online within the next few weeks, along with a recording and a transcript. I will send a link to the recording in the next day or so and will also notify all attendees once all materials are posted online.
Talking Freight seminars are eligible for 1.5 certification maintenance credits for AICP members. In order to obtain credit for today's seminar, you must have logged in with your first and last name or if you are attending with a group of people you must type your first and last name into the chat box.
Certificates of participation are also available for Talking Freight seminars. These certificates may be used for 1.5 professional development hours if accepted by your licensing agency. To receive a certificate, you will need to fill out a form. Please see the link in the chat box. Certificates will be emailed one week after the seminar. A seminar agenda has been included in the file download box for those who need to submit an agenda to their licensing agency.
Finally, I encourage everyone to please also download the evaluation form from the file share box and submit this form to me after you have filled it out.
Today we'll have three presenters:
Our first presentation will be given by Dr. Jose Holguin-Veras, the William H. Hart Professor and Director of the VREF Center of Excellence for Sustainable Urban Freight Systems at Rensselaer Polytechnic Institute. He is the recipient of numerous awards, including the 2013 White House's Transportation Champion of Change Award, and the 1996 Milton Pikarsky Memorial Award. His research interests are in freight transportation, and disaster response logistics. His research has led to major changes in transportation policy to improve urban freight systems. According to Google Scholar, he is the most widely published and cited freight researcher in the world. His work on disaster response has played an influential role in disaster response procedures and has led to deeper insight into how best to respond to large disasters and catastrophic events. He is in numerous leadership positions at professional organizations, public sector agencies, and leading journals.
Jose Holguin-Veras
Fantastic. It gives me great pleasure to talk about this exciting project of freight efficient land uses. Essentially, the goal of these procedures is to help maximize the benefits associated with freight activity, the production and consumption of goods. The main thing here is to facilitate the seamless integration of freight activities at all levels from producers all the way to receivers and into the fabric of the economy. This goal of seamlessly integrating freight activity is now made even more urgent due to the climate change crisis. In reality, cities and urban areas are manufacturing powerhouses that are accredited with producing tremendous amount of CO2 emissions.
Now, with the recent trends like e-commerce, there is an added pressure to increase the sustainability of freight activity. In this slide, I am presenting our estimates of the freight traffic measured by number of deliveries in traditional areas before e-commerce, with e-commerce, and with the impact of the changes in the economy that have taken place in the last year. As you can see now, in essence, the number of deliveries to cities have more than doubled and that adds tremendous pressure to the traffic network. And in this context, it is even more important than before to have a seamless integration of freight activity into the fabric of cities and metropolitan areas.
I'm going to start to try to answer the question, why do we need these concepts? I will show a couple of real-life examples. The first one is part of New York. At the beginning of the 20th century, the city of New York had one of the largest ports in the world. Over time, port activity gradually moved to the New Jersey side which created problems at the time nobody anticipated, because the reality is that a significant portion of the cargo arriving and destined to New York City was being delivered to the New Jersey side, but the bulk of the cargo was needed at the other side of the river. That basically put tremendous pressure because all this cargo had to move from New Jersey to the New York side using the limited number of bridges and tunnels that connect both sides. And that, in our estimation, has created at least a billion of dollars in congestion and externalities over the almost 80 years that have elapsed in the move to the New Jersey side. The chief insight of that is that when we do a land use planning policy, we need to take into account the impacts across the entire supply chains. It's not only the impact produced by a node, we had to take into account what are the implications of moving a node in the supply chain across the entire supply chain from producers all the way to consumers.
Another example is that is worthy of mention is that of the city of Cali in Columbia. Essentially, this city is located in a valley. To the left you have the Andes Mountains and then here to the right you have the Cauca River. Essentially, the city is sandwiched between the mountains and the river and the flood plains. Over time, the patterns of development that have taken place in the metropolitan area until the movement of manufacturing from centers to the north of the area while universities, retail, low income housing and the urban growth is going south. And now we have these kind of estranged pockets of development, freight activity to the north and to the south, urban dwellings, retail, et cetera. As a result, the flow of traffic is tremendous because people from the low income workers, that basically live in the south, have to travel north to get to the manufacturing sites. And then all the cargo that is manufactured for urban consumption has to travel the city. Basically, it's complete chaos. These are the things that happened when we don't take into account effective land use on freight activity.
Now, let me start with talking about the question, what is freight efficient land use? In essence, we define freight efficient land use as those patterns that minimize the social costs. That means the summation of the private cost of operating supply chains plus the externalities they produce. I need to make clear that this is an aspirational concept that is expected to be achieved in a gradual way as we improve the efficiency of land use. Now, key principles of freight use. The objective is to minimize social costs. Now, how do we achieve that? The first thing is to foster compactness of supply chains. In essence, the more compact supply chains are, the more efficient and also the less external [Indiscernible] we produce. Another important principle is to minimize the externalities that take place at all the stages of the supply chains. Another principal that is super important is to seek appropriate solutions to the context in the ways the supply chain is taking place. And also, finally, engaging the stakeholders is essential.
In order to illustrate these things, I am going to discuss three different alternative locations. Here you have a case in which you have a regional D.C., we have an urban D.C., and finally the [Indiscernible] to D.C. And basically, the goal is providing the idea about level of congestion; the red is more congestion than yellow. Now here, in this first location, this is bad for everyone, bad for the private sector, bad for the city. Because, again, the journey from the regional D.C. to the urban D.C. is long, you have to traverse the city, it makes no sense. It produces a lot of externalities, but also it is bad for the color. Now, if we have another location like this, in that particular case it will be better for the city, better for the environment, better for the private sector because the journey will be shorter. But here, in the location nearby the urban D.C., there will be externalities that will impact to the local community. Finally, we have another location like this that might indeed minimize the total social cost of activity, because by taking advantage of the bigger trucks to transport supplies to the urban D.C. it's going to use a total VMT. In both cases, in the second and third case, you have increased the amount of externalities that might affect local communities, that might produce community opposition. Now here, the fundamental insight is that if we address these externalities produced by the operations of the urban distribution center, and there's a way to do that, we are going to take significant steps toward what we call freight efficient land use or FELUs as we call them. The intent here is to explore synergies between land-use initiatives aiming at improving efficiency of supply change coupled with transportation initiatives to address issues that could be obstacles to the implementation of FELU ideas. In essence, the idea is to explore the power of both transportation policy and land-use policy.
Some of you may say, Jose, that cannot happen in complex, metropolitan areas. Guess what? Take a look at the case of Paris. The Paris metropolitan area is one of the largest in Europe and they have taken tremendous steps to fortify efficient land use. This slide shows some of the new terminals they are basically proposing. By the way, their objective is to use these terminals in combination with environmentally friendly modes. Just to give you a sense about the centrality of these activities, in these slides I am showing you the location of the Notre Dame Cathedral. In essence, these terminals are an example that we should all try to emulate.
As part of the projects, we produced a number of products. I will give you a very brief overview of them. The first one is the guide for FELU planning and decision-making that basically define the process. We also clearly defined the urban to rural transect and also provided a catalog of initiatives that should be used in order to increase efficiency of freight activities. In addition to facilitating implementation analysis, we developed a number of decision-support tools that provide assistance to the decision-makers and planners to take steps towards freight efficient land uses.
The first one is basically what we call the FELU urban to rural transect. It's a concept widely used that basically provides an idea or suggestion on how to integrate the different types of land uses. The only mention that we make about freight activity is in what I call special districts. These are basically the areas in which large manufacturers and the terminals alike are supposed to be located. It's basically our belief that instead of segregation for activity, we need to have seamless integration of freight activity. Why do we say that? Well, the reality is, notwithstanding the amount of traffic produced by large traffic generators such as manufacturing sites, and the like, the amount of freight trips they produce is minuscule, minuscule compared to the bulk of the freight traffic produced by urban deliveries. The restaurants in Manhattan, before the crisis, there were 10,000 restaurants produced about three times the amount of freight traffic produced by the whole of New York and New Jersey. And that is only one small segment. In essence, freight activity is pervasive. It takes place in households, retail locations, restaurants, offices et cetera et cetera. Segregation is simply not possible. Instead, what we recommend is to recognize, because in many cases some of these activities have already taken place. The idea is to somehow weave all of these activities into the fabric of the metropolitan area. That is what we recommend.
In addition, we produce an important catalog of FELU and transportation initiatives. In part of this catalog, this is basically some of the initiatives on the land use side. We catalog all possible ways to improve efficiency. Basically, all the way from planning, all the way to pricing, taxation, and incentives. Also, all of that should be supported by a vibrant process of stakeholder engagement. In essence, the plan here is to provide flexible paths that people could use to advance FELUs but taking into account the specific concept. The goal is to provide the [Indiscernible] with alternative paths to increase the efficiency of freight activity.
The catalog of initiatives is a superset of the initiatives we have identified as part of the NCFRP 33 that we conducted years ago, then we combined this with the one we developed for FELUs. We also included that into what we call an initiative selector. I suggest all of you take a look and try it. The FELU initiative selector is basically a dynamic database that takes advantage of the worldwide experience in the implementation of initiatives both land-use and transportation to provide the user with suggestions about possible solutions to the freight issues we are trying to deal with. Obviously, these take place in the meeting and planning, but the main goal is to mention ideas about potential solutions that might help you out.
As part of the initiative, the unique feature of this selector is that you specify a problem and then you get solutions. If you are looking for information about this initiative, you select them and then the selector is going to give you access to pages that summarize the key features, the pros and cons of the initiative that we are interested in exploring. In essence, instead of giving you a guide of 500 pages of material for you to find a solution, we have reversed the process. You define the problem and then we tell you where to look.
Another important product we developed is called the FASTGS, which is basically is a computer system that consolidates the literally thousands of freight models, freight trip generation models, and service trip generation models that we've estimated over the years used on the basis of the data that we have collected for almost 20 years. The unique feature of these models is that these models deliver establishment that by virtue of being an establishment level that allows us to aggregate them to any level of geography. We could aggregate results, buildings, senses, ZIP Codes et cetera. In essence, we have models in the thousands and every level at different levels of the industry sector. By the way, we also have a version that we could give free access to potential users.
Another tool we developed for the federal project is what we call the behavioral microsimulation. We have been developing this tool since 2005. This tool was to play a major role in the design of the off-hour deliveries project, that was basically the original use of this tool. Essentially, what we realized was using secondary data to provide an estimate of the freight flow that takes place in the city, exploring interconnections between the different economic sectors. It will give a sense about basically the total of the VMT that is being produced at a given metropolitan area. Also, it provides an interesting set of outputs for analysis.
Basically to conclude, the key goal of this presentation is to let you know about our work on freight efficient land use. The fundamental insight is that in an era of climate change we need to use all the tools that we have in order to improve the efficiency of supply chains with an eye to reduce the environmental footprint of these activities. The fundamental tenant of this project is that we need to explore these similarities between land-use and transportation policies. That doesn't happen often, by the way. Now, thanks to basically there is a number of efficient support tools. With that, I want to thank NCHRP, Dr. William Rogers and the project panel, they did a fantastic job at supporting us; the team at Rensselaer; Daniel Haake from HCR; The University at Albany, Cate Lawson; ATRI and Caliper for providing a tremendous support to us.
Jennifer Symoun
Thank you, Jose. I do encourage everybody to type any questions into the chat pod. We will address them after all the presentations. Please make sure you send questions to everyone. Our next presentation will be given by Dr. Monique Stinson, a Computational Transportation Scientist in the Vehicle and Mobility Systems Group of Argonne National Laboratory. In this role, Dr. Stinson leads studies in freight transportation systems analysis, examining the systemwide energy, emissions, and congestion effects of technologies ranging from e-commerce to electrified vehicle powertrains. Since 2010, she has led or contributed to agent-based freight model development efforts in Chicago, Arizona, and Singapore. Her ongoing work received Best Paper Award from the premier European agent-based modeling conference, ABMTRANS, in 2020. Dr. Stinson is a member of the TRB Freight Planning and Logistics Committee and a Handling Editor for the Transportation Research Record. You can go ahead.
Monique Stinson
Thank you, Jennifer. Good afternoon and good morning, everyone. I am delighted to be here to talk about the research in agent-based freight model development and applications we have been working on at Argonne National Lab. First, I would like to thank all the people that have contributed to this presentation in some way. Olcay Sahin, Vincent Freyermouth, and many others at Argonne; Amy Moore at Oak Ridge National Lab; Anna Spurlock and Victor Walker at Lawrence Berkeley and Idaho National Labs. Today I will talk about an overview of the transportation systems models that we have been developing. First I'll be talking about our flagship activity-based dynamic traffic assignment platform, POLARIS; then focusing on the agent-based freight model development; talking about some previous studies and then what's next in terms of their applications of this model.
The group I am in is the Vehicle and Mobility Systems, or VMS. Our tools give a streamlined way to conduct system-level energy efficient transportation research. We have tools: Autonomie which is used by hundreds of companies and auto makers to evaluate energy and emissions for single vehicle; corridor level analysis with a tool called Roadrunner; and then the software that I am mainly working with is called POLARIS where we simulate activities of all individual households and businesses in a metropolitan region for a span of one day.
These three software is our kind of at the center of this Department of Energy funded SMART 1.0 research workflow, which concluded last year. This was made up of a consortium of five National Labs. So, we had a lot of input related to EV charging, land-use, vehicle market, feeding into these models I mentioned and then the outputs are things like vehicle energy, mobility and energy productivity, and so on.
The tool I have been mentioning I mainly work in is called POLARIS. To date it has been under development for well over 10 years. To date it is focused mainly on passengers in the transportation system. It simulates their long-term choices around home and location, vehicle choice, their medium term location such as activities they want to pursue throughout the day, and then their activities during the day such as how to schedule those activities and route themselves through the transportation network. In the last three years, we added an e-commerce module to this to project the e-commerce demand for each household in the region as well as a so-called top-down freight and logistics model, which I'm currently converting to an agent-based model.
The agent-based freight model framework that I'm implementing is called CRISTAL, it stands for Collaborative Informed Strategic Trade Agents with Logistics. Basically, these are firms and establishments that form supply chain partnerships to trade goods with each other, they form a supply chain distribution network to transport those goods and distribute the goods through distribution centers and warehouses. And then the trucks they use form tours, and we simulate those tours in a traffic simulation. This was part of my dissertation and I would be happy to provide more information if people would like to see that. So, at the end of the day, we have all these passengers creating transportation demand in the system and then we have all of these freight agents. We keep track of the goods they carry, so we know what the payload of the trucks is as they move through the network and then we simulate all activity over a single day. There is a multimodal and international aspect where we allow the agents to perform supply chains and trade goods internationally. And then nationally, we run a multimodal mode choice where they choose between truck, water, rail, and air to transport goods nationally.
At the validation side we have a lot of validation data. We have been working with INRIX, ATRI, Census, and have obtained data from Census and another number of providers to validate our models. We run a number of performance measures and because we have such a high resolution model, this allows us to aggregate the result in any number of ways. We can look at how the system is performing for a single fleet. What is the fleet delay? What is their travel time? What is their percentage of their load in miles? What's the total cost of owning vehicles? We summarize energy consumption, carbon emissions, greenhouse gas and really look at how the system works for individual fleets, but then also how the system is functioning as a whole. What's the total Vehicle Miles Traveled and so on?
I want to touch on key insights we found from SMART 1.0. Again, this concluded last year and sometime later in the presentation I have a link if you want to see more about the full set of results. For the SMART 1.0 scenarios in this multi-lab consortium, we came up with a number of assumptions regarding travel in the baseline and in the future. Population growth, assumptions, land-use assumptions, fleet electrification assumptions, and so on. Then, behaviorally, especially in terms of freight and systemwide travel, we had assumptions related to assumptions on autonomous vehicle usage, the amount of sharing vehicles among passengers using Uber or Lyft. And then on the e-commerce side we looked at what would happen if e-commerce increased, if it tripled or even was five times the amount that it was in the base year.
Some of our key findings were that freight movement will become increasingly important due to increased light duty electrification and freight demand combined. These are results from the Chicago metropolitan region, I should mention, where freight consumes about 1/3 of transportation energy in the base year. In the future year, we are forecasting it would grow to about 50% of energy consumption. The reason is we all know light duty electrification is here already. A lot of people are able to buy electric vehicles. They are kind of cost-efficient and consumers are willing to pay extra for sustainable mode. But cost wise, it's not quite there yet with the trucks, there are not a lot of options. At the same time we expect, especially with the freight analysis framework and growth forecast, we expect freight demand will continue to increase. What this means is there is a huge opportunity space for improved energy efficiency for freight.
We also looked at greenhouse gas analysis. I won't go into too much detail on this one, but we found some really interesting simultaneous impacts of vehicle technology improvements as well as behavioral changes. For example, how passengers ultimately end up sharing versus owning autonomous vehicles., that has a lot of systemwide implications. In the high autonomous vehicle scenario, we are predicting there will be a lot more traffic in general and that has a lot of impacts on freight transportation.
And then in e-commerce, and our colleagues at Lawrence Berkeley led a study to conduct a survey called Whole Traveler, and they found one in seven weekly shopping events has been replaced or substituted by a delivery trip. So, there's just a huge question over the last several years of, what is the net effect of e-commerce? We all know most people are engaging in more e-commerce, buying goods online rather than going to the store, but at the same time, we know that parcel delivery trucks are not as efficient as the cars or walking that people are using to get to the store. So, the big question has been, these are really not apples to apples in terms of energy substitution, so what is the net effect? This was a critical data source along with the National Household Travel Survey, which also provided a lot of evidence that people are indeed reducing shopping trips, comparing 2009 to 2017. And we used that Whole Traveler survey data to develop an economic metric model which we used to predict which households in the region are engaging in e-commerce and how much is their demand. We implemented this model in the POLARIS platform along with parcel delivery truck models that our partners at Oak Ridge provided. And what we found is that as e-commerce increases, this is actually lowering the overall system of VMT and energy. This happens, obviously, when delivery tours are efficient. So, if you're talking about same day, express delivery where a truck is dropping off one package, going back, that's not efficient. But, looking at the tour paradigm, which is what the majority of service is right now, what we found is that current shopping trips are actually so inefficient that adding an extra delivery stop to an existing tour ends up swapping out about 7 or 8 miles of driving on average with just the .4 mile at the margin for this extra bit of the tour.
Bottom line is looking forward to the future or higher level of e-commerce is, we would expect a substantial decrease in total Vehicle Miles Traveled or VMT, as well as a decrease in energy consumption. So, here on the left is the baseline where blue shows the light-duty vehicle shopping, and the green is the medium duty parcel truck deliveries. So, you can look at the left baseline and then the middle and right are the future scenarios looking at quintupling of household deliveries, and we see just a dramatic drop in VMT and energy. Obviously, this is from, I think I mentioned year 2017. We all know e-commerce is growing steadily and sometimes dramatically year-to-year, so we do expect these results to be changing in terms of what we see as the number of deliveries. You can read more about this. We did a webinar series so this link is for the webinars as well as the capstone report if you want to read more or you can follow up with me.
I want to talk a little bit about our current studies. Some of these are contingent on funding; we are funded year-to-year. So, to start with our high resolution model, we have passengers driving around, we have fleets driving around, we have transit driving around in the system. This is all done at the agent-based level, so this gives us a lot of resolution to model the system and all of its complexities. For future scenarios we are looking at the impacts of a lot of systemwide factors like conductivity, charging, land-use, grid, as well as individual factors and I will talk a little bit about the freight specific scenarios now.
The first set of scenarios have to do with urban freight technologies. Really what we see is a densification of urban freight infrastructure at the last mile level. A lot more lockers, micro-hubs, warehouses getting closer to the city center, drones may be happening someday. And at the same time, e-commerce as we know continues to grow. The first set of scenarios have to do with what are the effects of this densification of the urban freight landscape in combination with the increase in e-commerce.
This is called the second set of scenarios, but in reality, it's highly related to the first set. In the second set we talk about urban electrification decisions. In particular, as deliveries and as warehousing and as those distribution points get closer to the city core, what's happening is smaller vehicles make sense from a business perspective. As I mentioned before, smaller vehicles make it easier to electrify that last mile. The point of this scenario is, what are the effects of electrified technologies for delivery? And in combination with these urban trend shipment points, what can we expect for energy and mobility impacts?
We also have an off-peak delivery scenario, and we are in the process of running this right now for the Chicago Metropolitan Agency for Planning (CMAP) Region. And with this, a lot of municipalities and cities have constraints with off-hours delivery due to noise, emissions; people don't want noisy, polluting trucks outside the windows when they are trying to sleep. It's an idea that has been talked about a lot by Jose and others. CMAP has been interested in it for over a decade. But the fact that electrification may be right around the corner for a lot of freight, maybe it's time we can see some of these local policies changing and permitting off-hours deliveries. Of course, there's the business model aspect, which Jose and his team at Rensselaer Polytechnic and others have done a great job elaborating on what kind of businesses can accommodate off-hours delivery with labor, for example. When are people on-site? Do they trust the supplier and so on? It's complex but it's something that may happen, and we are in the process of looking at that right now.
We have a couple of conductivity and automation scenarios looking especially routing controls. This would be one where a fleet manager tells his vehicles what kind of routes they can use as they make deliveries throughout the day. We also have some scenarios related to curbside impacts. Perhaps not at this moment. There's less traffic, maybe less curbside demand right now, but it's something that could be returning with all the sharing activity, with Uber and Lyft drop-offs and delivery, we do expect that to kind of bounce back and it would be good to look at.
We have another project with Cummins Inc., an engine manufacturer, where we're looking at not just urban now, but we're extending the model to long-haul and seeing what are the impacts of connectivity in automation as well as vehicle powertrain technologies on both the regional haul and long-haul freight efficiencies. This project has just gone started and we are extremely excited about it.
And then finally, we have a project to implement hydrogen fueling and EV fast charging for medium and heavy duty fleets in our model. We are especially excited because, you know, with the new administration, there's a lot more interest in hydrogen fueling. So, this is giving us the capability to start running some of these scenarios.
Lastly, I wanted to briefly mention there are a couple of notices of intent from the U.S. Department of Energy that they recently put out. One is called SuperTruck3; that is based on looking at systemwide impacts of freight, optimization, a whole number of factors that folks on this call may be interested in. And there is another one related to low greenhouse gas, which is more about deployments with different fleets. But we are always looking for stakeholders or partners to collaborate with and to tell us what scenarios are good to run and is anyone interested in a demo. If you want to read more there is a link. But definitely feel free to contact me, there is my email address, if you ever want to talk more. That's all I have. Thank you so much.
Jennifer Symoun
Thank you, Monique. Our final presentation will be given by Huajing Shi, Principal Data Scientist at the Port Authority of New York and New Jersey, where she helps the agency leverage the power of data science and machine learning to make informed decisions. She is passionate about cultivating data literacy through data visualization.
You can go ahead.
Huajing Shi
How do I bring up my presentation slide?
Jennifer Symoun
You should be seeing them right now.
Huajing Shi
I am seeing Monique's presentation.
Jennifer Symoun
I wonder if there is a delay. I just took it down and I do have your slides up right now.
Huajing Shi
Maybe I can refresh.
Jennifer Symoun
Let's try that.
Huajing Shi
It asks me if I want to leave.
Jennifer Symoun
You can try leaving and rejoining. If you want to start, I can advance slides for you.
Huajing Shi
Okay. Let me try that.
Hello, everyone. Thank you very much for this opportunity to present our visualization tool to real estate and truck activity. Although the title of this presentation includes both "Real Estate" and "Truck Activity", my focus will be mostly on truck activity. For New York and New Jersey Metropolitan region, trucks play a critical role in regional goods movement. They bring containers to our marine terminals, bring air cargo to our airports, and they generate significant amounts of traffic at our Hudson crossings.
It is obvious that we need to understand truck activities, to derive insights to inform business decisions. However, to make sense of truck activities, it is essential to have the data about the real estate associated with the truck activities. Most of the trucks at our facilities have origins and destinations at warehouses, distribution centers, trucking companies, or industrial sites. The fundamental questions about truck activities are always related to their origins and destinations, and the routes used from origins to destinations. The Port Authority of New York and New Jersey operates ports and airports as well as bridges and tunnels that serve as critical regional infrastructure links. Therefore, we are genuinely interested in the truck dynamics linking between commercial real estate, ports, airports, and bridges and tunnels.
What we need to know about the truck activities in our region boil down to the basic question of "where" and "when" the truck trips take place. Please bring up the video for slide number four. For instance, when a truck made one stop at our port, that stop could be one of the many stops it made along its cross-nation journey. The port visiting journey shown in this video could be very different in nature from another port visiting journey made by a local drayage truck. Even if two port visits have the same arrival and departure time on the port, they could be fundamentally different in terms of truck operations, and therefore they have different requirements for good service at our port. To help us understand the truck operations and how to serve our customers better, we need to get detailed information about the truck trip chains, such as the number of stops in a journey, dwell time of each stop, trip distance and trip duration. We also need to know the real estate information associated with each trip end.
Over the years, we have built up a comprehensive database of the activities of large trucks that visited our 28 county region at least once during the sampling time period. This 28 county region includes the area covered by NJTPA and NYMTC plus two Connecticut counties. This is the same area covered by the NYMTC Best Practice Model. This sampling method makes sure that we get the wholistic picture of truck activities in and out of our region. Sometimes people ask me "Which project do you use this data for?" I understand why people would ask this question based on the traditional way of doing transportation projects. We adopted this long term data strategy where data can be used for a wide variety of projects in the future, projects we don't even know yet at this point. Our activity data of large trucks ranges over 10 years, starting from 2009. The 2017 data set for example, has over 1 billion records. Each data record is a time stamped location read. Data wrangling is the most time-consuming task, but also the most critical step in making the data useful and meaningful.
The data was processed first to identify trip ends. And then trips associated with critical facilities were identified and labeled using geo-processing functions, so that we can analyze specific groups of trucks that visit critical facilities. Please play the video for slide six. Next, we want to identify the commercial real estate associated with trip origins and destinations. Later in this presentation, I will demonstrate how we use this real estate information in two of our projects. Many local government agencies maintain local business entity data and make them publicly accessible. We are in the process of assemble these public data sources into one integrated database. We find it too expensive to build a region wide, comprehensive commercial real estate database in one shot. Instead, we would build the real estate database in an accumulative way, one project at a time. For each project we worked on, we would collect the real estate information associated with the truck trips in that project.
There are many real estate data sources from both private and public sides. The basic data components we need are business location, company name, and business category. The New Jersey Department of Labor provides an employer database which only includes employers covered by New Jersey's unemployment insurance law. This data is free, but not up to date. The map on the right shows the locations of the businesses included in NJDOL database, and we can see that there are a lot of business locations not identified in this database. As for private data sources, CoStar is one of them. It provides very detailed information on commercial real estate. However, CoStar subscription is quite expensive, and it has very restrictive limits on the number of records that can be exported by each account within each month. Also, it may miss some businesses. For all those locations associated with the truck trips, but missing from the NJDOL and CoStar database, we turn to Google products to fill the gap. The map on the left shows all the business locations we could eventually identify. Google Places is a very comprehensive business entity database. It can be conveniently accessed using an API which could be free, depending on how much you use the service. The free tier API is good enough for our data projects.
The truck activity data combined with real estate data can answer many operational questions. We have talked about our truck activity and real estate database, and that's a lot of data. How could we make the data helpful to our colleagues who directly work in operations, to see, understand, and utilize the data? To realize the benefits of the data in helping make informed business decisions, we need to build a bridge to close the gap between the large database and the end users. The solution is a user interactive, web based platform for data visualization and exploration. This diagram shows the application development framework that brings the data to the users through web applications. A variety of applications are hosted on this platform to provide easy access to the data from anywhere, at any time. We want this platform to function as the GOOGLE of truck activities. The development is a continuous process, and we are adding new functions based on users' needs. We chose R Shiny for web application development. R Shiny is easy and flexible. It has gained its great popularity over the years.
The project demonstrated here is one of those data visualization applications developed at earlier times. Please play the video for slide number nine. The web applications developed back then provide information and functions similar to those available in traditional transportation studies. The query results are summarized at aggregate zone level. What I learned in visualization app development is that, developers should resist the urge to pack too much information or functions into one complex data product. Once my colleague asked me whether this app can calculate some type of statistics of zone based OD, and I knew this information is available somewhere in the app, but it took me quite a while to dig into the app and finally locate the web page that reports the requested information.
Our web applications have improved over the years by reducing the number of pre-specified assumptions regarding how users would like to see the data. Please bring the video for slide number 10. The newer applications are more focused on providing exploratory user experiences through dynamic interactions between users and data. Instead of summarizing the results at pre-defined zone level, the newer applications let users define a zone on the fly and all the statistics associated with that zone will be reported as a dynamic output. The newer apps make it easy to look into the details of each individual truck trip, and they also have much simpler interface design for easy navigation. The project demonstrated here was developed for the port gate hour extension initiative. To realize the benefit of extended gate hours at our port, we need to make sure that stakeholders of the port community are coordinated in operation hours. R Shiny provides superb capability in implementing interactive functions between data plots and maps, so that the update to the data plots and maps can be synchronized. For example, users can select a subset of the trucks based on the truck departure time from the port facility and display only the destination locations of those selected trucks on the map. The top origins and destinations can be identified in this app, and the operations time window at these business locations can be estimated using the truck arrival and departure times at those locations.
The web application demonstrated here was developed to estimate the potential Electric-truck conversion rate at our airports and ports. We need to estimate based on current travel pattern, what percentage of trucks are likely to convert to electric trucks with current and future battery technology.
We are particularly interested in trucks that are frequent visitors, as the conversion of these trucks would bring greater environmental benefits to our facilities compared to other trucks. Please bring up the video for slide 11, okay.
This application makes it easy to single out these frequent visiting trucks and find out their average daily travel distance which is then compared against the driving range on a single charge. We can use this app to explore the relationship between facility visiting frequency, trip length, and dwell times, which could help us identify the preferred locations for installing electric vehicle chargers, such as those locations visited by a large percent of facility visiting trucks. The truck stops in the map were color coded by dwell times, so we can easily identify stop clusters by dwell times. The electric-truck conversion rate could be higher, if we selectively install EV chargers at those locations where the average truck dwell times are long enough for a full charge.
This is an example to demonstrate how we integrate data with map to create interactive web applications using R Shiny. Please play the video of slide 12. This open source application uses the public dataset of Freight Analysis Framework. I think most of our audiences are familiar with FAF. FAF version 5 was just released recently, and it has year 2017 as its base year. I was wondering how the 2017 base year data in FAF version 5 compared to the forecast 2017 data available in FAF v4, so I developed this R Shiny app to compare FAF version 5 and FAF version 4 data by origin-destination pair.
You can access the data and R code of this app at my GitHub repository. I will also write the documentation for this application, and post it on my GitHub later this month. The link is not very clear.
In summary, from data engineering to web application development, then to application deployment on Azure Cloud, the projects I presented today were all developed in house, using only Port Authority resources. At the end of my presentation, I would like to give a big shout out to our research partner ATRI. As most of you may already know, ATRI are experts on GPS data analysis. That concludes my presentation. Thank you very much.
Jennifer Symoun
Thank you, Huajing. I know we have some questions typed in. Please feel free to continue typing in your questions as we go through them. Huajing, since you just ended I am going to go ahead and start with the two questions for you. The data mining can cause privacy concerns. What are data suppression or data protection issues?
Huajing Shi
First of all, this is a very valid concern, and we are very careful when we use the data. We are not going to disclose the details of each truck trip or do anything detrimental to the truck community. There is a rule for us when we got permission to use data is to improve operations, increase efficiency to help the trucking community. So, we are really very careful and aware of the issue.
Jennifer Symoun
Alright. And another question for you. Not all trucks will or can report locations. What expansion to all trucks are made?
Huajing Shi
That's right. The data we have is only a sample. For the trucks coming to our port facilities it roughly represents about 6% of them. In terms of informing business decisions, a sample with this size seems to be very useful and gives us very meaningful results. For the missing one, we don't know what we don't know, so it does have its limits but anything like if you use a data sample, you have to accept its limitations, but you need to be aware of that.
Jennifer Symoun
Alright, thank you. I will move on to some questions for Monique. I didn't see any questions for Jose. If you do have any for Jose, go ahead and type them in. Monique, we'll move to you, there are a few. Does your agent model address or minimize deadheading by freight vehicles?
Monique Stinson
That is a good question. Right now, we are in the process of developing a tour algorithm. There is some minimizing of deadheading in that. So, yes.
Jennifer Symoun
Alright, the next question. Why would I share my autonomous car? I don't share my Uber.
Monique Stinson
Yes, I realized actually as I was talking that maybe the way I was describing that was a little ambiguous. We have people that do participate in sharing, and then in the model there are people that own their household owned autonomous vehicles. So, today's paradigm where say Uber or Lyft central management and then private vehicles are autonomous. That being said, I have done some other research where we looked at whether people are interested in possibly sharing their autonomous vehicle. So that is, they would own the vehicle and it's autonomous and then they can kind of rent it out in a way to let other people take rides. There is some interest in that is what our study showed. If you want to follow up with me, I would be happy to share a paper on that.
Jennifer Symoun
Alright, thank you. Next question. What is the definition that you use for "e-commerce"?
Monique Stinson
We are focusing in this study on household e-commerce orders of retail goods. I know the census has in their surveys of businesses it's a little more broad. Their surveys will ask, for example, if a business basically makes orders by email or some other online ordering system that's considered e-commerce. But in this, we are focusing on delivery of packages to household, and so that's our definition of e-commerce
Jennifer Symoun
Another question. Did your model consider electrification of USPS trucks and other vehicles? And then the next question was, what are the assumptions regarding MHDV electrification? So, I think those two go together.
Monique Stinson
Ok, I will start with the first one and then maybe ask you again for the second. So, regarding USPS vehicles, we do have background traffic in the trip table, sort of a fixed demand trip table which includes USPS, it includes the cable company, all sorts of other light duty and medium duty type commercial traffic. We did apply some electrification factors to that pool, but not specifically to USPS, so they are covered through that. That is an area we would really like to get more detailed data so we can model trips better. What was the second part of that question, Jennifer?
Jennifer Symoun
What are the assumptions regarding MHDV electrification?
Monique Stinson
So, our assumptions regarding, I think that means medium duty, heavy duty electrification. I would have to revisit. They are laid out in great detail in those capstone reports. We looked at a short-term and long-term which are roughly 2025 and 2040. And then we looked at baseline technology, a business-as-usual case (BAU), and then what's called a Vehicle Technologies Office (VTO case), which is progressively more aggressive with assumptions. In terms of how things are looking even a couple of years ago, the prognosis for electrification adoption among freight was low. So, even the aggressive assumptions were on the lower side on the order of maybe 5% or 10% market penetration of those powertrain types. Again, the report spells it out in a lot more detail.
Jennifer Symoun
Alright, thank you. Will the models be updated to reflect California's regulation requiring manufacturers to produce zero-emission trucks? I believe other states are looking to replicate this regulation increasing the amount of electric trucks produced.
Monique Stinson
That's a great question. And the way we address the availability of electrified vehicles and regulations, you know, we have a few levers in the model that we use to address that. First, our firm and establishment agent go through a process of selecting a vehicle fleet. That's going to be based partly on what's available in the market. It's also going to be based on the total cost of ownership, and we have extremely, I mean, the research I present is on the systems side, but the total cost of ownership research is just as detailed and elaborate. The availability obviously comes into play and then total cost of ownership can include things like incentives, anything regulation-oriented that would have some kind of financial benefit. But then in terms of vehicle availability and charging availability, our model is also linked to the grid, so we can capture some of that. And then the fleet choice models are behavioral models. We did some light duty commercial vehicles preference estimations about maybe a year or two ago using data from the California Energy Commission. Our goal is to improve on those models and then apply those to medium duty and heavy-duty fleet ownership as well.
Jennifer Symoun
Thank you. And I should mention, all the presenters, feel free to jump in if there's anything you want to add to any of these questions. The next one might get some responses from the other presenters. Monique, I will put it to you first. One of the constraints of vehicle electrification is battery charging time. A delivery/charging model where "last mile" delivery vehicles operate during daytime hours and are charged overnight when the electric grid has plenty of capacity seems promising from the standpoint of grid capacity. But this would not allow for off-hour deliveries with these vehicles. Have any of the models and research efforts gone to this level of detail to incorporate considerations such as charging time, grid capacity, etc.?
Monique Stinson
We have done a lot of that research on the light-duty side linking our models to the grid. My colleague Omer Verbas presented a TRB on that. Our goal is really to bring that into, and it's in progress actually, we are bringing a lot of that work over to apply to medium duty, heavy duty now. In terms of what would be available for off-hours delivery, that's an excellent point. That is certainly something we would want to consider. Since our model is very high resolution, we actually model every six seconds of the day. So, we start with simulation at midnight or 5:00 A.M., whatever, run it throughout the day, so we have the temporal resolution to really reflect what does the business need to do in terms of delivery and what does it need to do in terms of charging. And yeah, the charging time, operationally I could see it working for a business that maybe charges the vehicle from, I don't know, 7:00 p.m. to 9:00 p.m. and goes out to make the delivery. We will also have extreme fast charging in the model over the next few months, so that is something else we could combine with an off-hours perspective and see the scenarios.
Jennifer Symoun
Alright. And before I move on, I didn't know if any other presenters had anything to add. Jose or Huajing, I didn't know if you had anything to add to that one. Alright, well, Jose and Monique, the tools look helpful. Thank you. The outputs of congestion and emissions are good, but can we add affordability? For example, the barrier to shipping by rail seems to be Class I rules favoring minimum shipping sizes (unit trains, full carloads), dumping higher costs on the very large number of smaller producers and shippers. How do we best examine the impact of such access or price discrimination on smaller producers and shippers? Monique, we will start with you since you have been responding and then Jose, we will go to you next.
Monique Stinson
That's a great question. One thing that we do in our model is we actually simulate all of the agents around the country. That's why business size, the industry sector that the business is, and their location, and then we simulate their trade partnerships. I would say we haven't, at this point yet, but looking at unit trains and say the associated costs with that, that is the kind of detail we can get into the model. Right now, we use a simpler tons-per-mile calculation. But we also model the shipment size, so a very large shipment traveling on a unit train is definitely something that we can do. In terms of looking at other modes and the price point that different sectors face, it is possible to look at something like that. Our transit modes actually are extremely detailed, so we could probably build upon that paradigm to do that analysis. That is an interesting question. I feel like that has a lot of policy aspects, so I'm interested to hear what Jose has to say about that as well.
Jose Holguin-Veras
Yes, basically, let me step back one step. I want to answer this question on the basis of the key results of the freight mode choice we recently completed for NCHRP. As part of this important freight mode choice that I believe has been the largest [Indiscernible] in the world, we interview shippers in the U.S.; the biggest of the big. And we asked them about what the factors were that they considered when making a freight mode choice decision, particularly rail versus truck. Basically, the almost unanimous answer was, rates and reliability. I was puzzled, and I asked them, what about travel time? They almost all said basically if delivery is reliable, if the service is reliable, we adjust to travel time. Now, fast forward a bit. As part of the quantitative part of the projects, we have access to the confidential micro commodity flow survey data. We estimated, literally hundreds of freight mode choice models for different industry, part of the country, etc. Low and behold, what was a key output of the models? That transit time was not a factor. Because, in essence, particularly for long-distance freight movements, they simply adjust their supply chains. The implications are, one, that trying to increase the market share of rail by means of reducing traffic times is bound to fail. That's the first implication. And the second implication is that in order for rail to increase market shares, increasing reliability is essential. My point is not only the cost. It's cost plus quality of service. Thanks.
Jennifer Symoun
Alright, thank you. Before I go on, Huajing, anything you want to add to that?
Huajing Shi
Not really.
Jennifer Symoun
Okay. José, a question for you. You spoke on the impacts of land use on freight patterns. A number of freight focused land use concepts such as Cargo Oriented Development, Freight Efficient Land Use and others have been developed. Do you have a suggestion on how to integrate these approaches into future urban planning curriculum, training, or outreach?
Jose Holguin-Veras
It is really important. It is important for universities, research groups, and the like to examine the dual effect of transportation activity and land use. Sometimes in most transportation programs, we simply do not pay attention to land use that we tend to take for granted. In our view, part of the analysis that we have conducted as part of this freight land use project to reveal the staggering toll of land-use patterns that are not conducive to freight efficiency. We collected, for instance, GPS data from a deliver company that is located in Pennsylvania looking to deliver to New York City. Basically, if we do not foster compact supply chains, we are doomed. Now, the only concept that you mentioned like cargo oriented developments and the like, they are all good things, but we need to do more on that. It's not only about freight activity doesn't take place exclusively at the locations of manufacturing sites, ports and the like. Freight activity takes place everywhere. Freight activity is pervasive. And now with the adding of e-commerce and B to C deliveries, even our households are basically receivers. In other words, we need make it easier for the manufacturers, the delivery companies, and the like to be embedded seamlessly in the fabric of metropolitan areas. What that means is that we need to take very seriously the effort of reducing or eliminating altogether the external produce on the communities surrounding the facilities. If we don't do that, there will be no political appetite for trying to implement freight efficient land uses. Thanks.
Jennifer Symoun
Thank you. Huajing, another question for you. For the dataset for your model, does the data include an O/D address? Are you able to identify the trucking company from the data?
Huajing Shi
No, the data from the raw format doesn't provide that kind of information. We need to process the data to first identify and differentiate the moving point from the stop point after identifying the trip end and associate that with real estate information. We need to integrate the original truck data with other real estate data sets to combine them, integrate them to figure out the business associated with the trip end. That is something quite sensitive and we don't really show that information to a very broad audience. It is limited access information after we process them.
Jennifer Symoun
Monique, another question for you. Do you have any ideas/suggestions on how shared rail infrastructure between passenger rail and freight might impact the models you are describing?
Monique Stinson
That is a really good question. A number of years ago, I guess 2008 or 2009 when I was a consultant with Cambridge Systematics, we did a study with the Chicago Metropolitan Agency for Planning looking at real crossing delays and share track and some of the issues, and single track, double track, triple track kinds of questions. It's something that we could do, but one impediment is the routing data from railroads is extremely proprietary information. If you look at the CREATE Project in Chicago, I think that's a good example of how they were able to share the data to kind of come up with some solutions. That's something I definitely would be interested, but it would have to be a matter of the data being available. Bruce Peterson at Oak Ridge National Lab has worked a lot on that data. I wouldn't rule it out and would be interested, but data permitting and stakeholders would help drive that kind of analysis.
Jennifer Symoun
I think we have gotten through all of the questions that have been typed in. I am not seeing anything else. We do have about one minute remaining. I am going to start closing out, but if I see additional questions, we can try to get those addressed.
I do want to thank all three presenters and everybody for attending today's seminar. I will send out a link to the recording of today's webinar within the next day. The April Talking freight is not yet available for registration but once it is an announcement will be sent through the Freight Planning LISTSERV. The Freight Planning LISTSERV is the primary means of sharing information about upcoming seminars. I also encourage you to join the LISTSERV if you have not already done so. The link is shown on your slide. I have also included the center number if you're going to obtain AICP CM credits and a link for a certificate of appreciation. Certificate will be mailed next Thursday. With that, I don't see any additional questions. We will go ahead and end. thank you, everybody, and enjoy the rest of your day.
