Jennifer Seplow:
Good afternoon or good morning to those of you on the West. Welcome to the Talking Freight Seminar Series. My name is Jennifer Seplow and I will moderate today's seminar. Today's topic is Freight Models: State of the Practice and Needs for Improvement. Please be advised that today's seminar is being recorded.
Today we'll have three speakers - Tianjia Tang of the Federal Highway Administration, Randy Curlee of the Oak Ridge National Laboratory and Luke Cheng of Citilabs.
Dr. Tianjia Tang joined the Federal Highway Administration's Freight office from FHWA's Resource Center in April this year. In the freight office, his main responsibility is to run the Freight Analysis Framework and the Freight Modeling Improvement Program.
At the resource center, Dr. Tang served in the position of transportation air quality specialist. His main focuses were urban travel demand modeling and motor vehicle emission modeling in conjunction with transportation conformity issues.
Dr. Tang has worked in the transportation industry in a wide range of positions including senior environmental specialist, project engineer, project manager, and program manager. He is a registered professional engineer in Georgia.
Dr. T. Randall (Randy) Curlee is a Distinguished Research Staff Member at Oak Ridge National Laboratory in Knoxville, Tennessee and an Adjunct Professor of Economics at the University of Tennessee. He received his Ph.D. in Economics from Purdue University and is the author of three books and more than 100 other publications. The focus of his research is on transportation, energy, and waste management. Working with the U.S. Army Corps of Engineers, Dr. Curlee led the development of the Ohio River Navigation Investment Model (ORNIM), which the Corps uses to assess optimal investments in new and improved inland navigation infrastructure. He and his team continue to work with the Corps Navigation Technologies Program (NETS), and he leads work for the Tennessee Department of Transportation to measure the economic benefits of public transit. Dr. Curlee also leads ORNL's support of the Federal Highway Administration's Freight Analysis Framework (FAF) and the Freight Model Improvement Program (FMIP), which promotes new freight analysis frameworks and models for all freight modes. Dr. Curlee's presentation today provides an overview of the Freight Model Improvement Program.
Luke Cheng has over 20 years of experience in transportation planning and traffic engineering, specializing in travel demand forecasting for passenger and freight systems. He is currently Regional Vice President - Asia Pacific for Citilabs Inc., a world leader in transportation planning software. Prior to this position, he served as Planning Manager for LA County MTA where he oversaw the development of a truck/freight movement forecasting model and provided support to other travel demand modeling activities. Before that he served as Director of Wilbur Smith Associates Limited in Hong Kong; Principal Associate for Wilbur Smith Associates in New Haven, CT.; Assistant Transportation Engineering for City of Upland, CA; and Assistant Transportation Planner for Nashua Regional Planning Commission in Nashua, NH.
I'd like to go over a few logistical details prior to starting the seminar. Today's seminar will last 90 minutes, with 60 minutes allocated for the speakers, and the final 30 minutes for audience Question and Answer. The Operator will give you instructions on how to ask a question over the phone during the Q&A period. However, if during the presentations you think of a question, you can type it into the smaller text box underneath the chat area on the lower right side of your screen. Please make sure you are typing in the thin text box and not the large white area. Presenters will be unable to answer your questions during their presentations, but I will use some of the questions typed into the chat box to start off the question and answer session in the last half hour of the seminar. Those questions that are not answered will be posted to the Freight Planning LISTSERV. The LISTSERV is an email list and is a great forum for the distribution of information and a place where you can post questions to find out what other subscribers have learned in the area of Freight Planning. If you have not already joined the LISTSERV, the web address at which you can register is provided on the slide on your screen.
Finally, I would like to remind you that this session is being recorded. A file containing the audio and the visual portion of this seminar will be posted to the Talking Freight Web site within the next week. Due to the size of the file, recorded files are available for viewing/listening purposes only and cannot be saved to your own computer. We encourage you to direct others in your office that may have not been able to attend this seminar to access the recorded seminar.
The PowerPoint presentations used during the seminar will also be available within the next week. I will notify all attendees of the availability of the PowerPoints, the recording, and a transcript of this seminar.
We are now going to wait a few minutes until 1:00 to give others a chance to join us. At 1:00 we'll start with the first presentation of the seminar. So, Operator, please put everyone back into hold at this time.
We've had a few others join us, so we're going to go ahead and get started. Today's topic for those of you who did just join us is Freight Models: State of the Practice and Needs for Improvement. If you think of questions during the presentation feel free to type them in the chat area. They will be addressed following all three presentations. Our first presentation of the day will be that of Tianjia Tang. Tianjia, you can go ahead and get started.
Tianjia Tang:
Good afternoon, good morning if you're on the west coast. I am with the Office of Freight Management and Operations here at the Federal Highway Administration in Washington, D.C. Before I start with my presentation, I will recommend two web sites to you. The first one is www.fmip.gov. This is an inter-agency effort among FHWA and the U.S. Army Corps of Engineers, USDA, and Oak Ridge national Lab, led by FHWA and operated by the Oakridge National Laboratory. The second one is the freight analysis framework side at www.ops.fhwa.dot.gov/freight. It's also known as FAF. Currently we are working on the second generation of FAF. I will give you another presentation on FAFII in the very near future.
Let's get back to my presentation overview. The outline of my presentation starts with modeling objectives, then modeling approach, from there it will cover the methods and will round it up by talking about the challenges we are facing. For us, modeling objectives, after extensive review and conversation with folks in the industry, it's very easy to conclude that freight modeling is centered around three main areas, they are economic development, logistic and supply chain and the transportation planning and engineering. It is mainly these subject areas we do freight modeling. There is a broad spectrum of the people in the organization in the freight modeling industry. Local state government agencies and federal agencies are all part of the process. It is very clear that freight modeling are not carried out by a single entity. My presentation will be focused on transportation planning and engineering, which is mainly the public domain.
Well, with the transportation planning, exactly what issues are we dealing with? According to the most recent survey, that is about 3 months ago, there are two major issues involving freight modeling. One is on the level where we're dealing with a group of projects, then issues associated with the projects, that is one project at a time. On the program level, freight modeling has been used in long range transportation plan and transportation improvement program. in areas where there are air quality issues, freight modeling has also been used in the conformity determination processes. Freight modeling has been used on the size and weight issue. On the project level, modeling has been used in aid of the design of bridges and roadways. To resolve a specific corridor issue. It's also been used in modal diversion such as highways to the rail. Freight modeling has also been used to locating weight stations, to optimize the location of parking and rest areas along the state highways. In the last couple years, freight modeling has also been used in studying Hazmat for roadway corridors.
The bottom line of freight modeling is trying to answer the following questions. First, the amount of freight goes into my jurisdiction, that means the amount of freight started from outside somewhere and goes inside. Typically refer the movement as external to internal freight movement. The second one is the freight goes out of my jurisdiction, however, this movement started within my jurisdiction. This is the internal to external movement. The third one is the amount of freight that's moved within my jurisdiction. Number four, the amount of freight moved through my jurisdiction that use our facility. That's referred to as internal and external movement. The last one is to determine the ways and methods the freight gets moved. Are they carried out by trucks? Rail? Water? Or air? What mode are we using there?
While we're dealing with freight movement and modeling, it's really important to realize the three different time lines we are dealing with. That is past, present and future. After all, it's really unlikely we are trying to solve yesterday's issue, unless yesterday's issue is repeating itself in the future. So we must pay close attention to the future data. Based on what I have reviewed, it is really really clear that a good modeling always has a very clear and concise objective. Clearly identifying modeling objective is a must. Also, as relates to freight modeling, we must be aware of short time planning by private sector versus long-term planning in the public sector.
Let us shift to a different gear and talk about freight modeling approach. We all know urban modeling, it has been practiced for over 25 years, and many consider it a mature science and art. There's a long history of urban modeling, it's for sure impacted our freight modeling. In reality, freight modeling has borrowed much of the skeleton of urban modeling, which is simulating behavior established mechanistic blocks and by using past and current empirical data to establish some coefficient to link those. Freight modeling is in large -- trying to simulate behavior. The four main blocks used by freight modeling are commodity production, commodity distribution, modes, and traffic assignment. Although there are exceptions . Those are the most used modes of the freight modeling process. Depending on criteria used, some people will come up with more methods, however, the criteria I use here is unique to freight modeling, basically, there are four methods, commodity based, trip generation, and the empirical statistical approach. The last one is trend analysis. The commodity based method generally follows the 4 blocks. It is also one of the most widely used methods. I will devote most of my time covering this method here. The commodity method started with commodity production, then goes to commodity distribution, from there we do the split for highway mode , once finished the split it will go to the vehicle trip generation, from there we will go to the traffic assignment.
Let's take a look at the production. We estimate the production consumption -- for example, in my area, how much alcoholic beverage do we produce? And how much alcoholic beverage have we consumed? The unit here is tonnage and dollar amount. The tonnage is good for loading the truck. The estimation of production and consumption does not need to be part of the model. There are a lot of economic models out of there, the key is to link production and the consumption to other economic, demographic data. Three key things we need to pay close attention to. Those three key things are commodity classification, geographic levels, and economic activities. The last one is the economic activity. When do you decide -- what did you use to tie those with commodity production. Commodity based modeling -- the next step is commodity distribution. The commodity distribution is to link production to consumption. That is to answer where our products are going to and where our consumptions are coming from. The distribution process establishes the so-called consumption-production pc data. How commodity distribution gets computed and allocated, it all depends on the type of data we have. Some types of data we do not need to distribute. The distribution may not need it. For other forms such as production consumption data you may need to do further distribution and when you are involved in distribution, the most widely used method is the gravity model .
Let's take a look at commodity flow survey data here. As you can see from the slide, we have production by area and column. We have consumption by area, and row. if we summarize a row, that is the consumption for total consumption for the area. If we summarize the columns that is total production for that area. For matrix like this, we may have missing values, if we have missing values and we are dealing with large areas, we commonly employ loglinear method. If you have marginal total data, can you also use the iterative proporation fitting method known as IPF process. This a slide that shows commodity production and consumption data. We have three columns here, the first one is area, the second one is production, the third is consumption. Clearly we don't know from this table here, we cannot tell where our production will go to and where our consumptions will come from . That is the case you will need, the distribution tools to distribute the production to the consumption. And get the consumption from the production. As I have said earlier, in the practical world, the gravity model, in one form or another, is the most commonly used method.
Now, we have commodity data, or PC data in hand. The next thing here is to answer the modal issue. That is to answer, how do commodities get moved from the origin to destination, by trains, trucks, water or air. Which mode are they taking? For mode determination we rely on surface transportation board's Rail data . For water, we rely on army corps of engineer's state-to-state information for waterborne. How about for trucks? We use what is leftover from the total. What about the model issue here, we have three different issues here, past, present, and future mode, the current data and the future. How do we handle the future model issues here? Typically we do it, for a given commodity, the mode remains unchanged. However, for the entire industry the mode percentage can still be changed. Now we have the mode data, we know how much is carried by the truck. For the truck parts, the next thing here, we convert all those tonnage data as I had said earlier, the production side is a tonnage and dollar, we convert the tonnage data to the number of trucks which is needed to carry those commodities around. So the key thing here is, to convert the tonnage to truckload. Where we get those factors? Well, they use to convert tonnage to truckload mainly from VIUS . The vehicle inventory and user survey. This is conducted by BTS of U.S. DOT. Local MPOs and state DOTs occasionally do their own conversion factors. I have a list of a couple examples by state and local agencies. The last step is traffic assignment. The traffic assignment procedures, virtually all the urban and statewide modeling used in traffic assignment for freight modeling, we must pay attention to the uniqueness of a freight shipment in the urban area with the injection of the logistics industry. Again, the size of our geographic coverage and resolution is critical in selecting this method.
Now we have covered all major mechanism blocks in commodity based modeling. What is the strength and what is the weakness of this method? The strength of this method is the commodity is linked with a lot of parameters. Economic parameters are there. The weakness is the difficulty you encounter of validation of this model. Think about it here, there are trucks on our highways carrying more things than standard commodity. As I said earlier, this method is the most popular one among all practitioners. You can read more of those types of modeling. The second most popular method is a so-called the trip-based method. It follows the trip generation, trip distribution , then goes to the traffic assignment, a three-step process. With the trip generation, it's most likely this relationship is built on economic activities and number of truck trips. For example, how many trucks will be generated for thousand square feet of commercial warehousing. There's no mode split here, we can go to the distribution directly. Again, the gravity routing model is the most widely used method. Once we have the truck number, then we can go back to the traffic assignment. Just like the commodity based method and urban modeling. Any of those assignment methods can be used, however, your geographic coverage and geographic resolution must be taken into consideration. The strength of this method is abundant l field data. Since this time, commodity definition is not a factor here. The definition for trucks is much broader, the weakness of this method is its inability to account for mode issue. ----
The next method here is the empirical statistical method. This approach is completely different from the traditional four-step modeling. This method basically is to establish an equation, a function which is the number of trucks, is a function of land use, business allocation and other information here. Let's take a look at this diagram here. We draw a line, and we say to start counting the number of truck traffic, then we call that number we have obtained as dependent variables. After we have the dependent variables, we go back to do more research to figure out what is the classification of the roadway and what kind of land use we have on the highway. And how far away is the business away from the counting station. We call all those whereabouts independent variables. We'll use the statistical method to link the dependent and independent variables into the equation, and we'll use that equation to predict the future data. Well, the strength of this method again. We have a long list classification and accounting data. The weakness is really hard to complete all independent variables. Most of those variables, so-called independent, they have some kind of relationship which make the issue much more difficult.
The last method I'm going to talk about is trend analysis. It's looking into what happened in the parts to obtain an equation, then we'll use the equation to project into the future here. This method is used on the price level. The classification number of the past, we'll use this regression into the future. The method is already seen put to use -- had been used extensively in the past. However, this method cannot be used on new alignments. Also, it will have a hard time to be applied on a widening job. The method is simple on growth rates and had been used a lot in the past, but in reality, this method is getting used less and less.
Now, we have covered all the methods, let's take a look what kind of challenges we are facing here. I call the first part a big challenge here, the modeling objective. I have stressed earlier, through reviewing all those papers and studies all those articles I have gone through. The modeling process must have a clear objective, if we don't have a clear objective, most likely it will not be a successful one. Relationships, we know there are a lot of predictions for future GDP's and if we can establish a GDP relationship with goods movement , then we will have a much better opportunity to project future shipment. The issue here is, the modal data. Right now we don't have a really good handle on how prices, markets, play into this mode of shipping. The last one is to integrate freight modeling software and other models, such as on the economic side the economic production model, and assignment side. combining with trip assignment . On the technical side, we need a way to collect and analyze the existing data. The algorithm is still the urban model one. We need new thinking of it. Freight has its uniqueness, has this injection of the logistics industry, how are we going to handle that. The last one is the model of operation and calibration. Let us recap what we have just gone through., we need objectives, there are a wide range of objectives we do in freight modeling, on the approach side, it's pretty much simulating our behavior. on type of models, we have four models, commodity based, trip based, statistical approach and the last one is trend analysis. Then we mentioned the challenges also. The challenges we talked about, the big picture and the technical side. I think I'm going to stop right now and thank you very much.
J. Seplow:
Thank you Tianjia and thanks to those of you who posted questions in the chat area, we'll get to those questions after all three presentations. I know he has some examples included in his presentation and web links and I think our other presenter do as well. These presentations will be available for download in the next few weeks and an E-mail will be sent out when they're available. You'll be able to get all that information then.
Our next presenter is Randy Curlee of the Oak Ridge National Laboratory. Randy, I'm just getting the presenter role turned over to you. Okay, everything's set up. You can begin when you're ready.
Randy Curlee:
Thank you, Jennifer. It's certainly a pleasure to be able to talk with people in the freight arena over this cyber net connection. Today I'm going to be talking about the Freight Model Improvement Program which Tianjia mentioned earlier, and more specifically about the clearinghouse that was recently opened at www.fmip.gov.
Let me first tell you a little bit about Oak Ridge National Laboratory. ORNL is the Department of Energy's largest science and energy laboratory. It employs about 4,000 scientists and engineers. ORNL also contains 18 large user facilities, one of which is the National Transportation Research Center which houses more than 200 transportation experts in a large facility in nearby Knoxville, Tennessee.
FMIP was launched back in 2004 by the U.S. Department of Transportation, managed by the Office of Freight Management and Operations. It's in collaboration with the Department of Agriculture, the Department of Energy, the Army Corps of Engineers and supported by the Oak Ridge National Laboratory. The FMIP is based upon three basic premises, the first being that the current methods for forecasting freight are inadequate; and I think that was one of the points that Tianjia brought out in his presentation. The second is that freight demand models typically are based on passenger travel demand forecasting, which has been developed extensively over the past several decades. But increasingly, we realize that freight is very different and will call for different approaches and different methodologies. The third point is that practitioners like many of you out there today are calling for improvements in both the state of the art and state of practice in modeling freight. So FMIP was initiated to do three basic things--to enhance the state of the art and the state of practice in freight modeling at the national, regional and local levels. The primary focus of FMIP is on forecasting movements of commodities by all the major modes and a secondary focus at this point of time is on models that estimate forecasts, public revenues, environmental consequences, economic consequences, et cetera, which is a secondary, but nonetheless important part of the program.
Many of you have asked about the relationship between FMIP and the Freight Analysis Framework which Tianjia also mentioned. The FMIP hopefully bridges the gap between FAF's national focus and the needs of local planning. FMIP is designed as a forum for practitioners, pretty much at the local level to help them understand -- help all of us understand -- freight through the further collection of local data and the modeling of that data as it is related to and dependent upon the national picture which is provided by FAF. A key component of FMIP is the website, and it has several objectives. The clearinghouse has been designed to provide up to date information on the current state of the art and state of practice freight modeling. It's designed to stimulate inventories and assessments of freight modeling. It's designed to develop and provide a library of current research data and methods. We are encouraging via www.fmip.gov, discussion and comments on issues, various issues that will help all of us lay out a path forward in the coming years as to what needs to be done to take us to that next level of freight modeling that we all acknowledge is needed. We are encouraging customers to submit copies of their work to this site so that we can disseminate that information more quickly and more fully, and we are also encouraging direct peer-to-peer exchanges through our site.
If you go -- and I hope that you all will go -- to the site, this is what you will see or something similar; It's updated daily, if not more so. You will find a welcome to the site. You will also at the right find what's new in FMIP, what's new in freight modeling, freight data collection. So I certainly encourage you to go to the site and explore it. If you look down at the major features of FMIP -- I'm just highlighting it here so you can read about it more easily -- we have news events and courses. We also have a section on freight models and modeling studies. We have a section on data to support freight models; a number of related links that are of interest; tools for those of you practitioners out there that might not know of these certain tools that you might use in your local, state, or whatever level of planning. There's the discussion forum that I'll come back and talk about later, and, of course, there is the contact section to lead you to folks who can hopefully answer your questions and respond to your comments.
If we go to the site and click on the data section, this is what comes up. Hopefully can you read that up at the top left, under the data section we have freight movements, commodity flow data, freight network data, economic activity, state and local sources and collection methods, I encourage you to go there and review these data. Time won't allow me to go into the details of that right now, but this is something I think is very valuable to all of us in the freight area. I do want to spend most of my time today on the highlighted feature here which is freight models and modeling studies and go through what is available there and encourage you to help us to provide additional information and make this an even better resource.
If you go to the site and click on -- under models, this is what comes up. Let me just highlight this and the next slide so it's more easily readable. FMIP under "models" has 5 other divisions. Under the first is an inventory of freight models, the second state and national freight modeling, and then a library of freight modeling papers. Several of you out there have inquired about the inventory of freight models. I have to admit that that inventory is still being finalized. As we go forward, I would encourage each of you to provide input to that inventory that we are developing and hope to have completed in the near future. If you go to the site, the page under national and international freight modeling studies, you'll see a number of references and links to work that's being done world wide from Australia to the United Kingdom to Sweden. And there are references to Japan. In fact we recently hosted a delegation from Japan. These scientists there are doing their own version of FAF, which Tianjia mentioned earlier. You'll see a number of links from California to Florida to Virginia. And if you click on these you'll get highlights of what those studies are, what they're objectives are, et cetera. You can also go to the metropolitan and local freight modeling studies; again, a large variety of contacts here and linkages from Baltimore to Kentucky to Los Angeles. I think these sites are very interesting in pointing out the different approaches being taken to different objectives of these studies which I'll come back and talk to a little bit later in my presentation. We also have a library of freight modeling papers, and it's currently divided into 17 different areas; and this is one of any number of ways that we can break down the papers, and the articles and the resources that are out there. Starting with freight modeling reviews -- and let me point out at this point that this library is designed for a very wide variety of users -- everything from the college student who is getting started in the freight area and wants to know something about the history and current state, to practitioners at all geographic levels, to hopefully the academic who might use the site for references for papers. So it covers hopefully most if not all the bases. We have sections on modeling hazardous freight movements, truck generation, modeling the value of time in freight, freight and commodity flow modeling, freight traffic assignment, modeling freight supply change, intermodal freight and transfer modeling, multistep freight modeling, freight corridor and system capacity modeling, dedicated truck lanes, truck size and weight modeling, modeling urban freight movements, freight mode choice, freight data collection, modeling the energy environmental impacts of statewide and regional freight planning studies. You could cut this any number of ways and I'll come back to this in the near future in terms of pointing out that this is one of the challenges that we have, in that it's not obvious how we do cut the current state of practice and state of excellence in this area.
One of the keys to success as we go forward in FMIP and hopefully go forward in terms of creating the next generation of freight models, I think it's first -- and this is to reiterate was Tianjia was stating earlier - that we understand the basic questions asked. What and how much freight will be shipped? We want to know the origin and destinations. We want to know what mode it will be moved by, or more appropriately modes, because we now live in a multimodal intermodal world. And I think we have been somewhat behind in acknowledging that in the models that we have. We want to know how it goes by route. We want to know the cost at which it travels, the transit time, and then finally, we want to know how these answers will change over time as there are structural changes in the economy and the infrastructure, et cetera, et cetera. It's important that we understand why we ask these basic questions. A number of reasons here -- safety, health, environmental impacts, national security, congestion. We want to know in many cases the damage of the freight movements to infrastructure. It certainly comes into urban planning and smart growth. We do these studies in large part because we want to better understand the relationship between freight transport and the economy at the local, state, regional and even international levels. And ultimately, we ask these questions, we do these studies, I believe, to assess the need for and viability of new and improved transportation infrastructure, again at all levels.
We have to recognize that our perception of freight modeling depends in part on our discipline, we must acknowledge that we bring different paradigms , understanding the problem in different ways, underlying theories, different assumptions in terminologies, depending upon the discipline in which we were trained; and these are facts that are both strengths and weaknesses as we go forward. I've just listed 5, 6 disciplines here, hopefully I didn't leave out one of yours; but if you look at these, we know the civil engineers approach this problem different than the geographer or the management scientist or the economist. And I think it's important that we recognize that. It also depends on our geographic scale of interests, whether it be local, state, regional, national or international. And on the role we play in assessing freight and, going-forward, whether it be as transportation analyst, an academic, a planner or the actual decision maker in both the private sector and the public sector.
I think improvements in freight modeling will depend on several things. First of all, we need to clearly articulate the objectives of models and studies which I think we've done a poor job of in the past. We need to adopt intermodal models and can no longer assess rail independent of truck. We have to play upon the strengths of our multidisciplinary collaborations, and develop a common terminology such that we can talk together more effectively. I think we have to draw upon our international friends and look for collaboration. The bottom line is that several countries are ahead of us in freight modeling, and I think there's a lot that we can learn from them. We have to wisely mesh together the theoretical with the empirical and the ideal with the practical. Modeling as we all know is as much art as it is science; and we have to be smart and wise in how we approach this. Finally, we have to make the most of our current data while laying a path forward for significantly improving our data, because our data are very much lacking in many areas, which Tianjia also pointed to.
In conclusion, may I first of all make a request that each of you go to FMIP.GOV and look at our forum. If you go to the site and click on discussion, this is what you will see. As you see, at the bottom there are discussions that you can click on. Currently we have 6 major topics of discussion. We've been a bit disappointed at how many people have used this, and we want to get this primed so that it is a point where we can start to discuss the major issues of the day and how we go forward in creating an agenda for improving freight models over the next years and decades. I also would strongly encourage you to provide input to us. We want your input and we want to serve your needs. To do that, please contact either Dr. Tang at the address on the screen, -- he, of course is at the Office of Freight Management and Operations -- and certainly feel free to contact me at ORNL. We certainly want to hear your comments. We want to meet your needs, so please, please come to the site and participate. And with that, I will end my presentation.
J. Seplow:
Thank you, Randy. I should mention that both Tianjia and Randy will be presenting again next year, I believe it's in February on the Freight Analysis Framework 2, and more information about registering for that seminar will be coming out soon.
Our final presentation of the day is that of Luke Cheng of Citilabs. So Luke let me just get you set up here, and you can go ahead when you're ready.
Luke Cheng:
I certainly appreciate being able to participate in this even though I'm on the other side of the globe here, thanks to modern day technology. And what I will be talking about today is basically what MTA, Los Angeles County Metropolitan Transportation Authority has been doing since 2001 in terms of trying to come up with a better tool to handle freight and truck planning issues in our region. So let's go on.
First of all, just a little bit about who MTA is or what MTA does. Most people know MTA as a transit operator, they operate a lot of buses, and they also operate metro rail system which include one subway line and three light rail lines, but in addition, MTA is also a county wide planning and programming agency, it is in that area that we have to deal with the planning issue and have to deal with the issue of freight and trucks. L.A. County has 10 million in population. In terms of modeling, we cover the entire southern California area that covers 5 counties with a population of 17 million about half the population of the state of California. That's a lot of people. And they consume a lot of goods. In addition, a lot of goods nowadays coming to the states, come through our ports, namely Port of Long Beach and the Port of Los Angeles. These two ports combined in 2003 stands at number 3 in the world in terms of container volume. In addition, we also have 6 freight and passenger airports, LAX being the biggest one and then Ontario airport. Also, and these are very active and busy airports in terms of cargo volume, and the two west coast class one railroads, Union Pacific and BSF are also very busy moving freight in our region. There are six intermodal terminals in our region. The biggest one, if I can get the pointer out here -- can you see this, the port is here, and the intermodal is right here downtown.
Everybody knows L.A. has a lot of freeways, most of these -- a lot of -- I don't want to say most, but many sections of these freeways are carrying over 20,000 container trucks a day, and the busiest one being the one connecting the port area and the downtown rail yard, the interstate 710 currently carrying about 40,000 to 50,000 container trucks a day. That's almost half of the volume of the entire road, and probably takes up 60, 70% of the capacity. That is now; if we look at the future, the population is forecast -- or expected to continue to grow, about 20% between year 2000 and 2020. For the same period, we're anticipating an even faster growth in terms of freight activities. Rail activities are expected to grow 240%. Trucks about 65%. Air seems relatively small in number, but in itself it's going to grow about 3 times. Since we're facing this problem, in our 2001 Long Range Transportation Plan, it was stated that MTA identified this as an important issue. MTA needs to more clearly define strategies to accommodate anticipated freight growth and MTA need to take a proactive role in working with all private and public stakeholders to develop some solutions. So this was handed down to the planning department and to the modeling group, namely the systems analysis and research section under the planning department. And the planners got together and I was hired to specifically deal with the freight issue.
First of all, just as Randy mentioned earlier, traditional travel demand modeling method is not suitable for modeling freight since the theory behind the methodology is how people are making trips. In the past, we tended to kind of ignore it, the way we dealt with it is by factoring the traffic assignment results up by 5% or 10% to account for truck. Now, there's a lot of traffic on our freeways and streets, that occupies a substantial share of infrastructure capacity, so we can no longer ignore this part of the traffic in our modeling. And as a planner, we need to understand better and get more knowledge in terms of how and why a freight truck moves.
So here is the MTA's objective -- our objective is quite ambitious, we want to develop an innovative, multimodal and comprehensive truck movement models. Innovative means thinking out of the box, just not constraining ourselves with how we model passengers. Multimodal - there is already a heavy duty truck model in our region; but that is not enough. We need to think multimodal. Rail and Air need to be considered too. Comprehensive - in an urban environment for an urban model, many trucks on our streets and our highways are not carrying freight. They're doing other things, they're providing services, so we need to take that into consideration also. So that's our objective. Coming up with a model that is innovative, multimodal and comprehensive.
After a year, we conducted many meetings, over 50 of them with different companies and agencies associated with freight industry. We also read a lot of papers and inventories of data, and then we sat down and said, okay, how do we model this complex system. In terms of modeling, the word modeling means, basically we're trying to simplify a complex system. So, okay, and I said very naively, let's just say all the traffic movement and freight movement in our region is split into three groups, one group is moving to and from the seaport, the other is to and from the airport, and then everything else got into another group - non port/airport related movements, these were to/from warehouse and distribution center, local truck trip, service oriented truck trip. And then maybe this model will look something like this. On the one side, we have a port and airport submodel which include two ports, six airports and six intermodal yards. On the other side of this is this non port/airport model. They will have submodel 1, 2, 3, but this part is a hard part, because nobody has really systematically looked at this yet. The port and airport related movement is comparatively simple because we're dealing with a limited number of trip generators and very luckily or fortunately in our region, Port of L.A. And Long Beach has done a great job developing their own transportation model already. They have a lot of data from the shippers and people doing business at the port and they conducted surveys on the trucks in and out of the ports. We don't have to reinvent the wheel. We can just ask them for their help and get their models over. On the Non port/airport related movement side, we're dealing with numerous origins and destinations. Even more so is that distinct type of operation. It's not like people making trips, these are business activities to allow these type of business to survive and make money -- like LTL, local delivery, service oriented, they're all different, the reason why they're out there is different and the operation is different, so they have different implications and impact on our infrastructure system. And also this part has not been analyzed systematically yet to date.
Two year after we started studying freight, we developed this document - Compendium of Truck/Freight Information for the Greater Los Angeles Metropolitan Area. -- the reason we put this together is that we have collected a lot of data and reports and past surveys and so on and so forth, we thought it would be a good idea to put it in one volume, a one stop reference for all the truck and freight related information for this region, and the idea being hopefully that 90% of the time when people want to look for some information, freight or truck related in the L.A. region, they'll be able to find it or find some thing that they can start with in this document. And it was proven quite useful for people in our region. It includes these 7 chapters, number 1 covers all the major freight transportation facilities, from the port to the rail yards or railroad company and airport, where they are, how they're doing in terms of movement and so on and so forth. We summarize all the freight movement data. There is some geocoding of this data and put a summary in this document. Truck generators, this section focused on where all these trucking industries are, and where all these warehouses are in the region. We produced some very nice looking maps, and summarizing this, the location of these generators. We often wonder how many trucks are there in our region? Does anybody know? So we started out with DMV data by getting inventory from DMV and summarizing it, and also we started categorizing all types of trucks and size of trucks and their gross vehicle weight. There is also a section on truck traffic; this started with Caltrans truck count database, and SCAGs, local truck count database, and also Caltran's OD survey. It also provides a summary of the major truck related studies in our region. Lastly, we also include a chapter summarizing truck involved or truck related distances in our region. I would encourage other regions to do the same. Once we did it once, then every year we can update it and it would be relatively simple to update. In the summertime, we can hire summer interns to spend a couple months on this to provide an annual update on this document.
The other thing we did early on was that we realized that we don't know freight. As a planner, we don't know very much about freight. So we needed some help. We hired a group of consultants to carry out a study that we call the truck freight modeling framework and preparation. The consultant team is led by freight industry experts, supplemented by traditional transportation planning modeling consultants. The objectives are number one to provide planners with knowledge and understanding of domestic truck freight movements, it's like a freight 101 for planners, if you will. And then based on that, develop a model framework for truck movement. And recommend an approach for domestic truck freight movement models.
The next few slides highlight some of the topics, which are covered in this report such as Trucking Fundamental, Freight Fundamental. There is also a chapter on local trucking. Local trucking accounts for most of the truck movements, not the long haul segment. There are three major types of local trucking operation. The radial one is the one we're most familiar with where you have a truck going out from the base in the morning to destination one and come back and then go out again to the destination two and come back. The other one is multi-leg, a little complicated, started out in the morning from the base, pick up the first load and go to the drop off point, drop off the first load, go to origin two, pick up the second load and then drop off the second load at destination 2 and so on and so forth. The third type is peddle. These will be like UPS or the post office. They start at the base and each truck will get an area. They drive out to that area first and then make the run. So these are description of how trucks move in terms of local deliveries. We also have a section on service trucking fundamentals, and most people didn't cover this. For a national model we probably don't have to talk about this too much, but in terms of local model, urban area model, this is something we cannot ignore. Service is a movement of trucks for the purpose of performing the service function. And this has rarely has been considered in planning or modeling efforts. Based on our study, 74% of the L.A. metro areas truck population is used in business or personal services. Most of these vehicles are relatively small in size but account for a significant portion of the vehicles fleet in our region.
Next we also have a chapter summarizing the state of the practice methodology in modeling freight movement and state of the art. Like I said, in the past, we just add say 5% across the entire network to account for trucks and then we go a little bit more sophisticated, say maybe freeways get 10%, arterials get 5%, and then another approach using OD based factors which means from A to B you get 10%. From C to D you get 5%. And now state of the art ones, the one that we picked out or the consultant summarized for us, one is a group called logistic chain models, and the other one is tour based models. At the end of the study, they recommended that a hybrid model framework will be the way to go. This will combine the logistic chain models and the tour base models. The logistic chain model basically is following commodity from the production to the consumption and this approach will probably be good for agriculture products, coal and mining industry. And the other methodology, the tour-based model will be suitable for textile apparel, electronics, furniture and also the service.
So just a few bullets of how each of these work. The logistic chain model, for example, is focused on how shipments move from producer to consumer, it includes mode choice decisions and three layers in this model component - economics, logistics, and transport. We estimate about 38% of the commodity can be modeled in this approach. And the tour-based model. These basically, like I said, for example, Best Buy they deliver their refrigerators to their consumers from the previous day's orders, so these basically are tours -- not your typical trip there's no mode choice since all tours will be made by truck. In fact some of the commodity they may have to be dealt with in both of these approaches. For the most part you will probably do it with logistic chain approach, once it gets into the urban area say into a warehouse, once coming out from the warehouse, they will have to deal with tour-based approach.
Based on that report and the recommendation, our next step is to conduct a prototype study, take one or two industries to follow through the logistic chain model approach or the tour-based model Since no one has ever done this we don't know whether it's going to work. We need to see what kind of assumptions we need to make, what kind of data are available and are collectible. And then see whether it can be worked out. Now, if it is successful, then the plan is to go on to the next phase to do it for the remaining industries, and then eventually do the model validation. That may take another 3 to 5 years. But in the meantime, we need to come up with a model that will allow us to forecast and estimate truck/freight movement. While the consultant was doing that study, internally, we were conducting our own research and surveys to see what other tools are out there that would allow us to come up with an interim model. We found Cube Cargo. Why Cube Cargo? The bullets on this slides are the justification we submitted to our management back in late 2003 for selecting this particular software. We are in no way endorsing any company or any particular product, these are basically what we did at MTA. After some research, we found that Cube Cargo, at that time it was the only modeling software available that is specifically developed to simulate regional and urban truck/freight movements. It was already proven successful and been used and applied in over 20 studies in various nations. And these range from a truck/freight model for city of Paris or the entire country of France and a national model for Germany or multinational model for along the Rhine River area,. Another reason for selecting Cube Cargo is, it meets our need in terms of our criteria number one is multimodal. It has trucks, rail and air modes, and inland waterway too if we have it. And it also considers beyond heavy duty trucks, it models local delivery, service truck trips in tour-based fashion, so that's another reason for picking this. It already contains logistic chain and tour-based concepts in there.
One particular module in Cube Cargo is called transport logistics node module, and this step is specifically designed to deal with the transfer issue, the intermodal yard, the warehouse where the goods get transferred, from the warehouse or distribution center. Lastly, for the reason for choosing Cube Cargo is that it's a module of Cube system of software, it's a family of modeling software, which integrates automobile, transit and truck freight into one system, at MTA, on the passenger side of the model we are already using Cube for the passenger side. Cube looks like this, it has an interface called Cube Base that is supported by ArcGIS. And it has a passenger travel demand model component called Cube Voyager, most of you are probably familiar with TRANPLAN, TP+ or TRIPS. Cube Voyager is a more up-to-date version of these software. MTA's passenger travel demand model is a TRANPLAN model which is part of Cube. So now with Cube Cargo that gives us truck and freight movement within the Cube system we can very easily incorporate that with our passenger travel demand model. Because a truck model cannot exist by itself. It has to be combined and looked at it in the totality of all the traffic out there. We also utilized another component of the system called Cube Dynasim, that is a software that allows us to cut out from our macro model a section or an area that we want to look into in greater detail.
For example, in our particular study, we utilized Cube Dynasim to extract out the I-710 corridor, the section between the port and the intermodal yard downtown and to develop a 3D Animation and micro simulation model for that section in order to allow us to compare different improvement alternatives for the section of the freeway. A lot of times we bought software but we don't know how to use it, we spent some money on the software, but ended up getting nothing. In this case, try to avoid that; in addition to just buying the software, we told the consultant we also would like you to provide a skeleton model for us, in a relatively short time, 2 to 3 months, using existing data. In this case, we can see the model structure, we can see what input is needed and what output we can get and what are the stages of it. What kind of assumptions and what kind of parameters we need. Then, after this step, we'll be able to know what additional works are needed in order to make this a calibrated model.
So at the end of the last fiscal year, we received these four deliverables. Number one, a Cube Cargo based preliminary truck/freight model. And a PC TRANPLAN and Cube Voyager version of our passenger travel demand model. In addition to that, we also received a GIS based highway and transit network to work with our macro models, and then lastly we zoomed into the I-710 area and develop this preliminary micro simulation model. Since last October or November we started phase 2, that is based on the phase 1 results, we started collecting data on intermodal yards and also on the I-710 and trying to come up with LA specific assumptions and parameters to calibrate this model. And that effort is still ongoing, hopefully at the end we will have a fairly decent model and I certainly think we will.
Finally, I think the goal for planners in the government agencies or consultants, is to develop a transportation system that is balanced. We want to balance the mobility for people and for the truck/freight. After past 30, 40 years of modeling passenger movement and transit movement, I think it's about time we also pay the same amount of attention if not more to freight movement, so we can have a balanced transportation system, thank you very much.
J. Seplow:
Thank you, Luke. We have a number of questions typed in, I'm going to try to get through as many as possible with the time that we have left. And just a reminder, unless you sent your question directly to me or to all participants, I am unable to see other questions that have been typed in, so if you didn't send it to me or all participants, go ahead and retype it and send it in to me so I can make sure I can see it. We're going to go ahead and start off with questions for Luke since he gave the last presentation.
Luke, the first question for you is how can one obtain access to the compendium of truck freight information?
L. Cheng:
You can send an E-mail to me and then I will refer you to the right person at MTA.
J. Seplow:
The slide that I have on the screen has the E-mail addresses of all three presenters on it. The next question is, does the truck compendium include studies from other groups such as SCAG or others?
L. Cheng:
Yes.
J. Seplow:
Why is freight growing so much more rapidly in Southern California than the population is? Is this due to the Los Angeles status as a major international Gateway? Or are there other significant growth factors?
L. Cheng:
That's certainly the major reason for it. And that statistic was prepared by SCAG. So I don't know what exactly went into that. I think the questioner is correct, it's mainly due to the influx of imports from Asia going into the United States and a lot of that volume, actually comes in from our port. And also, 17 million people consumes a lost goods, and we're consuming more and more goods that we don't realize, in fact with the advent of internet shopping, every time you click a mouse, you basically just attach -- dispatch a truck. And a lot of these trucks first may have to -- this growth may have to go by rail also. So all these additional consumption or addition of activities we don't realize we're causing a lot more trucks to be on the road.
J. Seplow:
Okay, well, thank you. I'm now going to move on to some questions for Tianjia. Where do you get the information regarding production and consumption amounts for a given commodity on a sub county area?
T. Tang:
Well, for production and consumption data, typically what I see folks doing is, by building correlation regression equations, based on past data, you know, tidal appreciate lace or something like that. Then for the future, that equation projects future production and consults. You use that consumption. The key thing there is what kind of parameters are on the highways, that's combative to the resolution, typically, people try to tie in with things in the future. If you can correlate the past production consumption with the population in some kind of equation, you have the future population projections and that is how people -- I don't see any of that in the literature, how folks are doing it.
J. Seplow:
Okay. Thank you. The next question I believe is for you, Tianjia as well. But if any of the other presenters have any thoughts, please feel free to give yours. How do you determine the total logistic supply chain movement from origin to destination across all model interchanges.
T. Tang:
Well, you know, our experience here basically, from the freight analysis framework here, we have gone through it once in the past, that has a summary that mainly dates from the former associates. And for the current freight analysis framework, what we did is we based it mainly on our data there from across the supply chain and based on the commodities. And then we conduct individual studies, try to supplement the commodity data with our own individual study. You know, for commodities which is not covered by commodities full survey. That is how we did across the board.
J. Seplow:
Okay. Randy or Luke, did you want to add anything?
R. Curlee and L. Cheng:
Nope.
No. Thank you.
J. Seplow:
The next question and again, I will pose it to Tianjia first, but Randy and Luke feel free to jump in. What is the best way for modeling freight flow at multimodal facilities?
T. Tang:
That is a tough question, but the key thing here, my take on it, the key is, it comes down to the model split, you know, what is model split. What is the facility? Those are the key things here, typically, what you design at the facilities, you'll have some kind of projection. So that is the key. What is the model -- mode of split there and what are the keys to facilitate. That is the way can you start.
R. Curlee:
I would just add to that that that's one of the major challenges that we have right now. And it's due in large part to just the lack of data. Luke has talked about the issues they have in Los Angeles and it's a very complicated issue, because we don't have those micro date that that really tell us what's going on at the level of detail that we need. And our hope is that over time the kinds of studies that Luke and other organizations out there, the other Metropolitan areas and regional studies are going to help us understand that better. And once we have the data, I think we can model it appropriately, but we really -- data poor right now.
L. Cheng:
I'd like to agree with Randy, and that's what we were trying to do at the local level, at least at MTA, so as part of the phase two for the Cube Cargo model, the main part of the work is actually trying to collect data at the 6 intermodal yards and we started off with thinking, you know, it's because I'm a traffic engineer, the only way I know how to do collecting data is stop people at the gate and ask them where they're coming from, where they're going to. And I got voted out by the consultant; they say that won't work. Eventually I gave in, so the approach that we eventually took was to go to the customers of these intermodal yards and I want to stress that -- the fact again is that it is good that we have freight consultants in our consultant teams, and these consultants are in the business, they know the freight people, they know the railroad companies, they know the customers of the railroad companies. So we go straight to the major customers, like probably 10, 12 of these companies which account for 80% of all the activities. We asked them for their movements, their trucks going in and out of those rail yards for the entire year of 2004. I think we were fairly successful in getting that data. Although we're less successful in getting the data from the railroad companies. So once the phase two study is it done, I'm sure MTA will be happy to share that with everybody.
R. Curlee:
If I could just add one point to Luke's comment. That and also Tianjia's earlier comment, our severe lack of information about truck movements. The -- we have to basically sub tract a way what we know about rail and water and deal with truck and inevitably we -- without collecting micro data, we're forced to use modeling approaches that give us reasonable estimates. But certainly that's something that you would say that particular movement that's estimated as in fact the movement that's occurring.
J. Seplow:
Okay. Thank you. The next question I'll pose this to all three of you is how reliable are the data available from commercial sources such as tram search? Randy, if you want to go first?
R. Curlee:
I really, I don't feel qualified to answer that question, I know it's certainly been used and we have used the data in the past for FATH, and the folks that prepare these data are very, very high quality folks and so I would suspect that the data are very good. We currently don't have the complete background and information on the methodologies that are being used. So we can't do a complete evaluation of those data, but there's no reason to suspect that they are not anything but first rate.
L. Cheng:
I would have to agree with Randy, I don't feel qualified to comment on this, but as the user or practitioner on the government side, my feeling is it's better than nothing, a lot of effort went into it, and like Randy said, a lot of people's efforts on it, it also has a long history. For the last few months, I've been running around in Asia, and I realized in the U.S. we're quite fortunate. At least we have something to start with or compare against. At this point I just wanted to say one point that I forgot to mention in this presentation. I want to thank FHWA for providing the highway and rail networks and all the commodity flow data as part of the Cube cargo model effort. And also Cal trans ITMS data these are commodity flow data at the county level and Zip code level. That's a big effort they put in. And that benefited us. After we develop the model, we can compare our commodity flow at least on the Zip code or the county level against ITMS data. So I want to just say that to give the credit to these two agencies.
J. Seplow:
Thank you. The next question is after converting commodity tonnage to trucks, most freight planning studies use all or nothing assignments, assuming no passenger car traffic. Only trucks on the highway network, how do you account for congestion? If any one of you feels like you can answer that, please jump in and go ahead.
L. Cheng:
Well, I guess I'll start, like I mentioned, this is Luke, freight cannot just run by itself on the road, so any reasonable modeling effort, once you have the truck matrix, you need to combine that with the passenger car before you do the traffic assignments, somehow you need to account for that in order to get the congestion.
T. Tang:
It's all conjecture on that analysis. We look at it as a combination of fact. It cannot be a truck alone or -- it must be a combination. That comes back to objective study. I want to say one thing, the search data, as most of you folks already know, Randy said, the freight framework, the first one, generation one is mainly based on transportation data. What we're working on now is a framework for the second generation, we shifted to two different sources, but that has nothing to do with reliability of numbers, so we change the course of the data. It's due to other reasons.
R. Curlee:
Let me just add that the problem of truck converting tons into truck is one of the issues we looked at early on in the process. We have a working paper that I'm sure will be available for public consumption that's done here by Dr. Frank south worth who's one of my colleagues here and looked into the truck payload equivalent issue. It's not an easy one, again, it comes back to having the viable data, to transfer if the tons and the truck payload equivalents and there's a hole science that deals with that, but I think it's one of those sub issues that we have to throw into the pot as we go forward to develop the next generation of freight models. It's very important to not only understanding congestion but also damage to road infrastructure, et cetera.
J. Seplow:
Okay. Thank you. We have just enough time to ask one more question, and then if I didn't get to your question or if you have additional questions, you can post them to the freight planning list serve and you can contact the presenters directly, I have their E-mail addresses on the screen. The final question I'm going to ask is that in general and from experience, getting shippers and in a large measure, manufacturers to participate in surveys that are either government mandated nor are voluntary is difficult. Are there ways to get through these issues and practice so the surveys are received well or replaced by other data sources?
T. Tang:
Well, that is a really tough question. It's not this unique toward the freight, you know it data service. Any other public service, that is always a challenging issue. I think the key here, for a local agency, you just need to work without a private firm. The builder of the trucks, that's the key thing there. Then I think, you know, it will be much easier to move the process forward.
R. Curlee:
May I just add that the next generation FAF is very dependent upon the commodity flow survey which is a shipper survey. And you have to recognize the limitations and the pros -- the pros and cons of that, we don't have a full picture of who is actually making the decisions about freight movements and that's something I think again we have to address as we go forward, what decisions the shippers make, but then we have to infer from that many decisions by other parties that we don't directly measure, so I think the question is an excellent one and one that needs a great deal of attention.
J. Seplow:
Thanks. Luke, did you have any thoughts on that?
L. Cheng:
Okay. Yeah, just very briefly, I think survey certainly is a big issue, and most of the surveys in the freight side are less than successful I'll say, and we ended up losing a lot of money. So coming up with another approach will be quite useful. But in the meantime, I think in order to improve the survey quality and response, I think the mentality on the government side needs to change also so we cannot just conduct a survey on the project by project basis, where we hire consultants to conduct surveys and get nothing back. We need to change the practice into a more institutional way of doing things, meaning like for example in the SCAG area, they have a goods movement task force, this is a permanent group that meets every month that involves the freight industry people or the key people there. So when the government has needs, you can discuss in this forum and you have trucking industry people there, trucking association, railroad company, they're all there in the same room and so these trucking industry people and shippers even, they will go back and convey the government's approach and the intention. So they can help us in terms of the study we need to do and the data we need to collect. That's how we were able to collect this data in the intermodal yards and some of the major warehouses also. We need to think permanent, institutional and long term, in forming these task forces in each of our nation's different regions.
J. Seplow:
At this point in the interest of time I think we're going to close the seminar. I believe I got through most of the questions, but again, feel free to contact the presenters or send it to the LISTSERV if you do have additional questions. I want to thank all three presenters, especially Luke who is calling in from Beijing in the middle of the night there or early morning, whichever you prefer. Thank you to everybody for attending today's seminar. Again, the recorded version of the seminar will be available within the next week or so on the Talking Freight website. The presentations will be available for download as well, and I'll be sending an E-mail out to everybody who is in attendance to let you know when those will be ready. The next seminar will be held on December 14th and is titled "Considerations of Freight in Disaster Planning". Please note that this seminar is being held on the second Wednesday of the month due to the holiday season. If you have not done so already, I encourage you to visit the Talking Freight website and sign up for the seminar, and I also encourage you to join the freight planning LISTSERV. One more thing I want to the add, the first few seminars of 2006 will be posted to the Talking Freight website in the next few weeks. As I mentioned previously in February, there will be a seminar on the Freight Analysis Framework 2 in which Randy and Tianjia will be presenting. That's about it, so keep an eye out for the seminars for 2006 and you can begin registering for those as soon as they're posted. Thank you, everybody, and enjoy your day.
