David Anderson:
I believe we should go ahead ... we should go ahead and get started. Good afternoon, everyone. Thank-you for joining us for our webinar. As we get started I am obligated to read a disclaimer to you all: This Webex call is being recorded and it may be posted on DOE's website or used internally. If you do not wish to have your voice recorded, please do not speak during the call. If you do not wish to have your image recorded, please turn off your camera or participate by phone. If you speak during the call or if you use a video connection, you are presumed to consent to recording and use of your voice or image. Now, that may not impact many of you because you should mostly be muted from the host end, but we have to cover our bases nonetheless.

Alright, we can go ahead and get started. If we can pass control to Aymeric to present the slides ... and Aymeric, if you can pass me control to ... control them ...

First slide:
OK, again, good afternoon, everyone. I am very excited to welcome you all to our first in a series of webinars to highlight some of the work, some of the results, the findings and insights, from DOE's Smart Mobility Lab Consortium. My name is David Anderson. I'm the program manager for Energy-Efficient Mobility Systems, or EEMS. That's a program of the Vehicle Technologies Office, part of the Department of Energy's Office of Energy Efficiency and Renewable Energy. The Smart Mobility Consortium is a cornerstone research effort of the EEMS program, and it's focused on understanding the system-level impacts that emerging transportation technologies and services will have on mobility, and on identifying solutions that improve mobility energy productivity. That's the energy, the time, and the affordability of transportation. Our plan is to have webinars every two weeks from now through early December, each one focusing on a different aspect of what we accomplished and what we learned through the first phase of smart mobility over the last three to four years. Today our focus is going to be on the smart mobility modeling workflow; that's a core capability that we developed focused on large-scale transportation system modeling and simulation. Before I turn it over to Aymeric to get into the details of the modeling workflow, I want to give you a little background on the consortium to motivate the discussion and to provide context for what we're doing. ... I don't seem to have control of the slides ...

Next slide:
Alright, so when we think about it, mobility is the foundation of how humanity interacts and connects, and that's not an overstatement. It's driven the development of cities and of civilizations for centuries. Transportation gives us access to opportunity in the form of healthy food, good jobs, quality education, adequate health care, and plenty of leisure and recreational activities. And with the current state of technology advancement, we have the potential to not just move more, but to move smarter. Now, smarter means more affordably, more efficiently, and with more choice. The understanding and knowledge that we're creating through the consortium and the solutions that we develop will help us usher in the next transportation revolution, one of smart mobility.

Next slide:
So why did we decide to undertake this challenge at this moment in time? Well, we did it for several reasons. Our transportation system is being influenced by several important factors, among them are a growing and aging population. Our population in the U.S. is expected to grow by 70 million people in the next 20 years. We also have increasing density in our cities and our urban cores. Every week one and a half million people move into our urban areas. This population growth leads to more congestion on our roads, more pollution in the air, and more greenhouse gas emissions, impacting our climate. We have 3 trillion passenger vehicle miles traveled, and 11 billion tons of freight being moved through our roadways annually. They all tend to want to move at about the same time each day, and so this results in congestion that costs the average American driver $1,300 per year in lost time, adding up to a drain of $300 billion dollars to the U.S. economy. In addition to these economic drains, there's also the direct cost to travelers. Transportation is the second highest cost for U.S. households; only housing is ahead of it. U.S. households spend an average of 20 percent of their income on transportation, so despite these challenges our transportation system is fundamental to our way of life. All the things I talked about before -- our jobs, our health care, our education, our recreation; business is accessing markets, our nation's economic productivity -- is all made possible by the transportation system. It is foundational to the way of life that we enjoy.

Next slide:
So at DOE we have a mission to accelerate the science and technology that drive U.S. prosperity and security, and we have a critical role in realizing the potential of smart mobility. Across our national lab complex we have unparalleled computational tools, scientific and engineering talent, and R&D resources that deliver impact by creating solutions that address our nation's complex long-term challenges. And so we decided to take a multidisciplinary consortium approach to transportation, because the opportunity space for mobility is much larger than any one organization. And we convened five of our leading national labs doing transportation research, including dozens of talented transportation scientists and engineers, to build the Smart Mobility Lab Consortium, a multiyear, multilaboratory collaborative dedicated to further understanding the energy implications and opportunities of advanced mobility solutions.

Next slide:
Now to pause just for a bit, we use the term mobility quite a lot so it's important that we agree on what it means. If we want to improve mobility, how do we know if we're successful? We surveyed the transportation landscape and found there were no existing mobility metrics that serve our needs. Current transportation metrics tend to focus on utilization of the road network, but we're not just interested in improving the road network. We're taking a multimodal approach to connect people to the things that they need better, while at the same time minimizing energy consumption. So we define mobility to mean the quality of a network or system to connect people to goods, services, and employment, the things that define a high quality of life. And we defined a metric called mobility energy productivity to quantify mobility, and that will be the topic of a later webinar.

Next slide:
As I mentioned earlier, EEMS is one of the research programs within DOE's Vehicle Technologies Office, or VTO. Now, many of you are likely aware of the R&D that VTO traditionally does, shown on the left side of the screen, focused on vehicle power trains and component technologies. Now, with EEMS and smart mobility, we really extend VTO's reach to the systems level, to the right side of the screen. We look not only at connectivity and automation at the vehicle and component level but also at how vehicles interact with each other. And with infrastructure at the small network or corridor level. We can then evaluate the impact that they have on overall traffic flow and energy consumption. And finally, we can understand their implication in the context of an entire urban or metropolitan area including both passengers and freight and not just on-road vehicles but also transit and micro-mobility and pedestrians and biking -- the entire mobility system.

Next slide:
So our consortium has concentrated on five focus areas to conduct its multidisciplinary research. The first is connected and automated vehicles. This is where we identify the energy, technology, and usage implications of connected and automated solutions. Secondly, in mobility decision science, we work to understand the human role in the mobility system, including decision-making and technology adoption. Next, in multimodal freight, we evaluate the evolution of freight movement and the impacts of new modes for both long-distance goods transport and last mile delivery. Fourthly, in urban science, we work to understand the linkages between the transportation network and the built environment. And finally, in advanced fueling infrastructure, here we try to understand the cost, the benefits, and the requirements for fueling and charging infrastructure to support future mobility systems.

Next slide:
So the challenge is to cut across each of these research pillars and consider transportation holistically, that is, as a system, highly interactive, highly dependent, very complex system of systems in which changes to one factor propagate and result in changes to all the other components. And so while we have to understand each of these systems individually, it's just as important or even more so to understand how they fit together, how they interact, and how they affect one another.

Next slide:
For example, as new automated driving features are introduced, we have to understand how they operate -- not only how they impact the vehicles equipped with them, but also how they affect the surrounding traffic.

Next slide:
As new mobility services reach maturity, what does that mean for how and how much people travel, the number of cars on the road, how we manage the curb space?

Next slide:
In addition to these new services, what about new modes? Is micro-mobility a viable urban transportation option, or will it be relegated to a niche role in just a few locations, and if it's the former, how will it impact other modes such as transit?

Next slide:
And finally, e-commerce. Now, this is a big one. We've seen a big shift from brick and mortar retailing to online shopping and delivery over the past decade, even before the impacts of COVID-19. And so to what extent will this trend continue? What impacts will it have on our transportation system, and are new and future modes like package delivery drones efficient? Are they affordable are they even viable?

Next slide:
So to answer these questions and to tackle this complex problem space, through the smart mobility consortium, we created an end-to-end multifidelity modeling workflow that really advances the state of the art and transportation system modeling and simulation. Ihis is a software tool chain built around agent-based transportation system models that capture the complex interactions among mobility decision-making, technology implementation, different mobility service models and modes, and land use, and EV charging infrastructure. It integrates freight and passenger movement, and it allows our researchers to evaluate the energy, the mobility, and the affordability outcomes of potential future transportation scenarios. And we can evaluate these outcomes in terms of a variety of metrics, including vehicle miles traveled, travel time, energy, costs, GHG and others. This is a really critical research capability, but we think the real value is in working with stakeholders -- the transportation planners, the technology providers, the policy makers and NPOs. Our plan going forward is to ultimately deploy this capability and put it in the hands of decision-makers, and we're already doing that. We're using these tools to help cities evaluate what a return to normal looks like for transportation in a post-COVID scenario.

Next slide:
And so now I'm going to turn it over to Aymeric Rousseau from the Argonne National Lab to describe how this capability was built, how it works, and to give a few examples of what it can do. And when he's done, we'll have some Q&A, and I'll remind you to please type your questions into the Q&A box in the Webex. And as time permits, we'll address as many questions as we can get to. Alright, I will turn it over to you now, Aymeric.

Aymeric Rousseau:
Thank-you, David. I'm Aymeric Rousseau. I manage the Vehicle and Mobility Systems group at Argonne, and today it is my pleasure to discuss the creation, and of the small mobility modeling workflow.

Next slide:
When we are looking at transportation, as David mentioned, there is a lot of technologies that are affecting mobility and energy. And so the research questions are numerous and diverse. So for example, you know, will this new technology and service promote or inhibit a shift of traveler, right, to for example passenger car to transit? What is the impact of the economized travels? What is impact of urban planning with increased vehicle electrification? You know, what, what happens to charging and fully fueling infrastructure needs with connectivity being more and more important? You know, what happens, for example, with traffic control infrastructure? How can we actually leverage that new technology? There is a lot of focus right now on passenger movement, but freight movement is also critical, especially with significant increasing e-commerce. So what does that mean? And then, you know, if people use different modes such as transit or ride-hailing, you know, what is the impact of private vehicle ownership? As David mentioned, you know, some of the many, of the core competence of the U.S. Vehicle Technologies Office resides in component and powertrain. So when we are looking at vehicle electrification, and we're looking at advanced engine transmission lightweighting, you know, what, what does that mean in a connected and automated vehicle context? So all of these questions, when we think about it and take them individually, they are very complex and require unique set of expertise and tools. But the key, as David mentioned, and the key challenge is actually to integrate them together because they actually are all interrelated. The increase in vehicle electrification is linked to charging and filling infrastructure. The fact that, you know, we have traffic control infrastructure, well, only this can be leveraged if a lot of vehicles are also connected and automated, and so on and so forth, right? These are the reasons behind the development of the end-to-end smart mobility modeling workflow.

Next slide:
What we want to do and what we've done is actually to integrate multiple high-fidelity tools to take into account the entire system, so that we can for the first time simulate the impact of different technologies simultaneously, looking at the system of system. At the center of the smart modeling workflow is the agent-based transportation system modeling. This is where we use mesoscopic simulation. So looking at a very large number of people, let's say, you know, 4 to 10 million people over a 24-hour period. And we can actually represent how traveler and goods move as well as the different options on the system control. But as we mentioned, this is not sufficient. We need to be able to understand how the entire system works, and for that, on the left-hand side, if you look at EV charging, right? So where are charging locations, where are charging stations located, what type of charging station do we have, what happens to land use? David mentioned that, you know, cities are expected to become, you know, more and more populated, but what does that e-commerce? We now have two-hour delivery; what does that mean? And then also in the vehicle market, looking at today's fleet distribution, right, to mostly distribution with the reductification and changing in vehicle class, that will also significantly impact the transportation system. So we need to take all of that into account. Now in addition to the environments, with EV charging, land use and markets, we also need to take into account the impact of technology. And that's where the multivehicle control and microscopic traffic flow simulations are very critical. At the multivehicle level, what we do is we simulate one to 10 vehicles, and we can actually predict how the vehicle will behave with knowledge of the environment, whether it's through connectivity or through automation, and impact the vehicles around them. So think about vehicle-to-vehicle, vehicle to infrastructure. So using this lesson to learn, and the small vehicle number with high-fidelity vehicle, part-time models, in control, we can then use that information to microscopic traffic flow simulation. Now think about 10,000 vehicles. And what we really want to do here is look at the impact of different market penetration of connected automated vehicle technology, and what does that do to traffic flow? And usually here we represent a stretch of highway or urban area with 10 to 15 traffic lights. Now that information from microscopic traffic flow can then be used and scaled up again to the mesoscopic traffic flow simulations. For now we're not just looking at the impact of this connecting automated vehicle technology on a stretch of highway or a stretch of urban, but throughout the entire system, right? So this allows us to simulate the entire system from all different angles.

Now, obviously what we want to do then is exercise this system, and David mentioned there is a lot of outputs that are coming out of it because we do have a lot of different parameters. One of them is obviously the vehicle energy consumption emission and cost. The way this is done, DOE has invested a lot of funding over the last 20 years to be able to simulate any type of vehicle partnering configurations across vehicle classes, from light to medium and heavy duty. And so to be able to properly estimate the energy impact of different technology now and in the future, we need to be able to recreate vehicle 3 profiles. So the mesoscopic traffic flow simulation will output an average speed over a specific link. So let's say in between two traffic stops. But as we all know, acceleration and deceleration are very, very important to energy consumption. So by smoothing the vehicle speed choice potentially through connectivity and automation technology, we need to be able to properly reproduce that and estimate energy. Now there's a lot of other metrics, and we're looking at mobility, you know, as well as many others. One of them David mentioned that we will be discussing a little bit later in this presentation, as well as in the follow-up seminar, is the MEP mobility energy productivity matrix.

Next slide:
So how did we do that? Smart 1.0 was a three-plus-year effort, and the workflow development was definitely a very large team effort across all five national laboratories. And I would mention later many universities also involved in the process. And it all started by doing modeling. We had more than 50 researchers across all different backgrounds come and meet together to try to understand what does make sense to link and what information should be provided from one model to another, so on and so forth. So very quickly within three months we were actually to develop a smart workflow proof of concept and so we use ACC and connected multiple tools to see and demonstrate that we could actually improve the fidelity of the different tools and the overall resonance. So now that we were comfortable and confident that we could do that on the small examples, we decided to scale it up, and we developed a gun chart across the entire construction as well as develop common scenarios you will see later. There is even if the small mobility workflow is generic, there is actually two implementation of that workflow, one led by Argonne National Lab, the other one led by Lawrence Berkeley National Lab. We also wanted to make sure that the workflow was generic enough, so not just two different implementations of the workflow but that within each workflow we could use different fidelity of different steps. So one way to do that was to develop and implement processes for mobile input and output so we could, you know, let's say, for micro-simulation either use Aimsun or vcm or sumo, right, regardless of the tools that are available. That they could actually be plugged into that workflow. And then we started exercising, you know, within 12 months. So very short timeframe between the first meeting and the actual implementation. And we were able to quantify the energy mobility and mobility energy productivity impact of new transportation technology. Right now, we're happy, and you know today is the first presentation to say that we do have a seamless workflow that leverage multiple expertise across multiple research organizations both in the us and actually worldwide.

Next slide:
So let's go through some examples and benefits as to, you know, what can we do now with that workflow and how does that work. The first one is really learning from detail model to scale to larger ones. At the end, as David mentioned, what we want to do is we want to simulate a very large number of people moving in an area, a mesoscopic traffic clock simulation. But to be able to do that, we need to have a good understanding of the impact of technology. For example, on the traffic flow. So the way we did that is we can start with a low level. So Road Runner here is used to develop advanced control that are enabled by knowledge of the environment, so connected and automation, and then once we learn a little bit how future vehicle could potentially interact. So think about a vehicle gap. We can then scale that up into micro-simulation. So now we're able to not just have two three or four vehicles interacting together but now thousands of vehicles with different market penetration. And what that allows us to do is to develop fundamental diagrams. As you can see here, one of one input that is critical to mesoscopic simulation here.

Next slide:
So one example here, because there's actually many many different cases that were run out of the micro-simulation, is looking at a fundamental diagram and some outputs. And when we take this individual output for different market penetration we can then generate inputs to the mesoscopic flow.

Next slide:
Another example is to try to maximize model reusability across projects, right? We know we all like to develop our own models, but we are a lot more efficient and consistent when we use models across multiple studies and projects. So our one example here is reusing autonomy for vehicle energy consumption, performance and cost. Autonomy has been developed under the DOE Vehicle Technologies Office over the last 20 years, and, you know, the model has been validated pretty much across the world. So we didn't want to develop new ones; we wanted to make sure that we reused it. And in addition, by using the same models across multiple projects, we're also able to compare the impact of technologies within these projects at a higher level. So as an example here, Lawrence Berkeley Lab used the autonomy for ACC and CACC control that was in micro-simulation outreach, also use it for CAVs coordination and Polaris, also used the same models for vehicle consumption of the entire system. So this is really allowing us to provide consistent and comparable results.

Next slide:
If you remember a little bit the high-level workflow, there was a lot of arrows and a lot of them were going back and forth. And that's because the fact that we have multiple tools, many of them require an iterative process, it knows it's not just the one flow where one model will provide an output to the next and you go on and on and on, and you have your results. And the main reason is that many of these systems, as we discussed, they are related. There is interactions. So one example is between land use and transportation system. So depending on the way people travel, depending on whether you will own a car or not, right, this will affect where you will live. So there is a clear relationship between land use and transportation systems. So this is an example of closed loop iterations between Polaris and UrbanSim that have been developed by Lawrence Berkeley National Lab. So we're looking at employment by industry, as input to Polaris, for example, and, you know, Polaris provides information back to UrbanSim as to the OD Skims by modern time building performance. So depending again on where people live and how they travel is very much interrelated. So that closed loop is critical for the overall workflow simulation.

Next slide:
Another example of closed loop is plug-in electric vehicle charging location and behavior, right? Depending on how many electric vehicles you have or plug-in hybrid electric vehicle you have, depending on where they are, depending on how they are used, whether they're personally owned or fleet-owned, right, this will significantly impact the number and type of charging location. In that particular example, NREL has been developed for many years now software called EVI-Pro that is really focusing on charger location and time. So that software was actually linked with the transportation system simulation, and Polaris here would do the vehicle routing, so where you are, where you're going. But we now have also integrated the machine learning model embedded within Polaris to estimate the energy consumption, that is, the state of charge of the battery. Now knowing the state of charge of the battery for an electric vehicle operating habit is not sufficient; we also need to understand when and how people make charging decisions. And so Idaho National Lab has collected over the years many data that were actually very helpful to develop this charging decision algorithms. The reason why this is an iterative process is that you know we would start with Polaris, and let's say, 10 market penetration of electric vehicle, we would let them run through the system, and people would decide when they would want to charge. And we would allow them to charge at will, right? This first pass will then be provided to EVI-Pro, and EVI-Pro will close the loop and say based on that particular demand throughout the day, this is what I believe the charging network should be and the type of charging location. And so we would close the loop again, and say now, we do have constraints on the charge locations, are we meeting the demand? It's all about supply and demand, in terms of when do I want to charge and where and can I do that?

Next slide:
As an example of outputs of the charging station, so baseline would be today within the Chicago metro area, as to where are charges right now, at home as well as public. And then we can look at many, many scenarios. The one on the right-hand side is a long-term scenario, with significant penetration of battery electric vehicle that are, you know, including lower cost and so on and so forth. We have a significant penetration of PEVs which drives quite a lot the number of required charging stations both at home and outside of home.

Next slide:
Now, the last benefit and that I will discuss today -- again, there's many more; these are highlights -- is of the small big workflow, is we can also have improved scenario. And let's go through the example of freight. One thing we're looking at freight is the impact of e-commerce and how e-commerce is becoming more and more prevalent over time, you know, even the fork of it but even some more now. What we started with is Lawrence Berkeley National Labs has been performing a survey called the whole traveler. And this again would be, I believe, the third webinar of the series. That survey data was instrumental to develop an e-commerce behavioral model. So who, which households will order e-commerce, you know, depending on how many people are in the household, depending on your occupation, depending on your finance, so on and so forth. Based on the e-commerce behavior model, we can actually predict the demand. Now from there, Oak Ridge National Lab, who has a lot of expertise into freight and has been working on delivery tools for many years, was able to develop different delivery tools. Based on, you know, where the deliveries are made, you know, how do we actually perform the delivery? And then this information was then provided into Polaris to develop the route network. So now we actually assign buses to specific routes with specific deliveries, you know. Let's say, taking an average of 120 deliveries per day for different trucks, and then we look back into the overall workflow by saying, OK, now we know where the trucks are, we know what type of truck we have, what are the energy and emissions related to these particular deliveries? So we use SVTrip for the vehicle trip profile and then Autonomie for energy consumption and emissions.

Next slide:
So there is a lot of models, right? We've named probably, you know, half a dozen so far. And then there is obviously more of them. And with these models we also need to do control; that's one piece we haven't discussed yet. When you're looking at all the high-fidelity models combined together along with control, we definitely need to make sure that we have an acceptable simulation time. And for that, we've been leveraging high-performance computing, especially for optimization and control in Smart 1.0, and will significantly be expanding this in future research. So when we're looking at high-performance computing, let's take two examples. The first one in the bottom is personally owned automated vehicles. So think about driverless vehicles. And we need to look at whether in that particular example, can one household that owns two vehicles replace both of them by a single driverless vehicle? So person 1 and 2 have their own activities, right? Person 1 goes to work and goes shopping. Person 2 goes to run some errands. When we are looking at, you know, one vehicle for two people, we may need to shift some activities, right? But we don't want to shift activities too much. But we can shift some activities. Let's say I will run my errands maybe a little bit later in the morning. So we update the schedule, and then what happens is that we can actually have a single automated vehicle for that entire household that will serve the needs of both persons. So the reason why we need high human computing here is that we have millions of households within a model, right? It's not just one time; it's multiple times. The other example is platoon formation decision. In that particular example, we have about 10,000 trucks providing deliveries in a large metropolitan area. The question was like, whose trucks can platoon with what other trucks and when should they leaf platinum, right? You don't want to join a platoon for a couple of, you know, for 400-500 feet, you won't want to join the platoon for a longer time, and you don't want to wait more than a certain number of time to join a platform, right? So opportunistic platoon versus planned platoon -- what does that do? Obviously here a very complex optimization problem, and high-performance computing is key to implementing that one.

Next slide:
I will show in the last couple of slides some results, and again, brush over them because all the following webinars will go into a lot more detail. But just to give you a glance, and some of the key outputs that the workflow generates is vehicle, my travels, and in some cases, you know, if we make moving cheaper, right, the people will tend to drive more. What is the impact of technology on the overall vehicle miles traveled, or VMT? We also looked at productive miles traveled, which is the total amount of personal miles traveled by all travelers in any modes. And that includes freight. That's really a measure of the load on the system from a user's perspective. And we also are looking at vehicle hours traveled. So we want to make sure that, yeah, people travel in long distances, but you know, how long do they actually spend on the road traveling? Productive hours traveled, or PHT, is the total travel time for all users in the system. And then we have couple of ratios. The average vehicle travel or network speeds, that gives a good idea on congestion, the average trip or travel speed, also, you know, important as metrics. And then one that starts to take I would say a higher level because it also does include energy is a travel efficiency. So in that case, like how much energy does it take to move a single person or good per mile. Now, we could list many, many more metrics, right? And in some cases, the commercial will go up, they could hardly travel, go down, and so on and so forth, and it starts to be difficult to understand what is the overall impact on the system.

Next slide:
And that's where MEP, mobility energy productivity, metric comes from. That metric was developed to quantify the energy cost and time weighted opportunity space within the reachable area. So if you live within a specific area, what are your options in terms of modes? And then, you know, what is the energy related to that? What's the travel time? What's the cost? And it actually weights all of these parameters into a single metric that we can then compare across different scenarios more easily. And as David mentioned, this will be the focus of the following webinar in two weeks. That metric was developed by NREL.

Next slide:
We have a generic workflow, and then here we wanted to make sure that as part of that workflow, right, the workflow was applicable to multiple tools. So we actually have done two implementations. The first workflow implementation was centered around Polaris and led by Argonne National Lab. Within that workflow, Road Runner was used for the multivehicle simulation. So again looking at advanced control, vehicle speed, vehicle speed and Powertrain control enabled by connectivity and automation, Aimsun was used for the microscopic traffic flow simulation. As I mentioned, Polaris for the transportation system simulation, EV charging was EVI-Pro, and land use UrbanSim. When we take all of this into account, we use Autonomie for vehicle energy consumption. And because the dynamic of the vehicle speed profile are critical to energy, we used SVTrip, which is a stochastic tree profile generator. When we're looking at this particular implementation, you know, all the models across the board here, and all the inputs, included all five national laboratories, but it also included seven universities, both in the U.S. and outside of the U.S. So Texas A&M, as you can see in the bottom, University of New South Wales in Sydney, Australia, George Mason University, University of Texas at Austin, University of Illinois at Chicago, University of Washington, and UC-Irvine. And the idea here is that not just reaching out within national lab systems to experts on specific areas, basic tools, but experts across the world. You know, for example, looking at value of travel time is something that University of New South Wales in Sydney has done for decades. You know, same thing on travel behavior. You know, University of Illinois, looking at TNC and ride-hailing, and many universities across, whether they are fully automated or partially automated. So we really looked at who was the experts in a particular field, and then rather than redevelop, we partner with the experts across the board to integrate their state-of-the-art models or data.

Next slide:
The second workflow was centered around Beam, and Beam is developed by Lawrence Berkeley National Lab. Beam and Polaris are very different; obviously we want to take different approach. That being said, there are some similarities within the workflow. For example, tools like UrbanSim for urban planning, EVI-Pro for charging station locations, or Aimsun for microscopic simulations. We are consistent across the board because we wanted to, you know, avoid having too many differences, as well as MEP for the overall results. But the structure of the workflow and, you know, how many additional tools are linked to it, again, very different. One objective here was to make sure that we could take two different approaches with different tools, some of them in common, obviously, and the workflow will still be fully functional.

Next slide:
So let's look at some examples, the first one starting at the individual vehicle.

Next slide:
We'll go from individual vehicle to entire metropolitan area. The first result was that we were actually able to demonstrate up to 20 percent of individual fuel consumption reduction through advanced vehicle enforcement control. And this is enabled by connectivity and automation. So obviously, the scenarios matter greatly when you're looking at energy. So whether you are in a suburban environment, as you can see, whether you drive a conventional or an electric vehicle, whether you have V2i or not, communication, all of these are very important. And we looked at all the different scenarios to quantify their impacts.

Next slide:
We also looked at the impact of different technology penetration to impact traffic flow. The first one was ACC, automotive cruise control, and what we've seen is that actually the lack of communication under some high penetrations can lead to significant traffic and instability and congestion, which as a result could lead to fuel consumption increase as much as 60 percent. So definitely, you know, we think this technology under some conditions has great potential, but it also has potentially great, great negative consequences.

Next slide:
So now we moved on to CACC, operator adaptive cruise control, and in that case we actually saw significant improvement on traffic flow, which as a consequence lowered the energy use up to 20 percent. Again, this is some high-level results. We'll go into a lot more detail, including the impact of the technology with different market penetrations.

Next slide:
So coming back at the mesoscopic level and the overall transportation system, so we have a lot of models; we can run lots of scenarios, right? And this is actually what we plan on doing definitely, but so far what we've done is consider three main scenarios. The first one, on the left-hand side, that we call scenario A, is shorter-term. Consider here that we have high sharing, so high ride-hailing, for example -- Uber and Lyft, with partial automation. And partial automation being introduced primarily on the highway. There is also small increase in e-commerce and vehicle rectification. But to a great extent, think about it as, you know, a high-sharing scenario, shorter term. The other two scenarios in high automation are longer-term. Both of those include some level of high-automation technology penetration, and that means in that case driverless vehicle. There is also obviously some partial automation, automatic vehicle here. But for the sake of simplicity, think about high level of automation. For both of these, we do see a private ownership change, as well as an increase in e-commerce and a vehicle rectification penetration. Now, the main difference between both of these scenarios is actually how are these vehicles used? On scenario B, those vehicles are used mostly by fleet. So think about ride-hailing, right, Uber or Lyft, using driverless vehicles, where on scenario C, those drivers vehicle now are used by you and I, private households. So in that case, you know, we would see that the industry has done a great job. Driverless vehicles are very cheap, you know, and many of us can actually afford them. So it's very similar technology in DNC, really looking at different usage. So let's look at some of the results.

Next slide:
And the first one is applied to Chicago, and that's looking at the impact of using these vehicles. The highly automated vehicle scenario with a lot of electrification with high sharing. So fleet on the left and personally owned on the right. What we see is that there actually could be significant benefits, even if, you know, we have a lot more people. Because it's long-term scenarios, population increase, everything, but we could see significant savings in energy -- of 23 percent, in that case. Benefits in terms of VMT, no change in PMT, and also a benefit in terms of increasing speed 17 percent, which means a lot less congestion. Now, using again this same technology but in a different aspect, now everybody, a large number of population, most of them are owned by private households. We see a significant increase in energy, 22 percent, increasing VMT by 25, and that's really driven by a lot of vehicles driving around empty. Now, that being said, we see benefits in PMT: 7 percent. But overall everything is at the expense of significantly increased congestion, with a decrease in average speed on the network of 18.

Next slide:
We talk about passenger car. I also wanted to show you an example related to freight. And this one is a really good example to show the synergies and why having a system of system is very important. When we're looking at increasing e-commerce, it really means two things. It means that we do have an increase in delivery trips, right, because we order more at home. But also as a consequence, we have a reduction in individual, the people driving to do shopping trips. So when we're looking at what type of goods are ordered, with e-commerce, actually the average shopping trip that is replacing seven to eight miles each way. Where delivery trade we have one at this top that is 0.4 miles on average, so obviously trucks on a per-distance basis consume more energy. But the added distance is much smaller. And what we see, and again, this is longer-term, we also have improvements in vehicle powertrain electrification, both for light-duty as well as medium-capability. We actually see potentially significant benefits in terms of the overall system vapor mass travels, as well as the overall system energy from 30 to 55 percent, depending on the vehicle pattern considered and the penetration of the car electrification.

Next slide:
So in summary, the small mobility workflow really provides researchers, you know, within the consortium, but as David mentioned, also outside the construction, because we do want to deploy this capability, unique set of opportunity to quantify the impact of advanced technologies at the system of system level, right? And so today we just glanced through a couple of examples, a high-level overview. David will mention that there is much more detailed reports that are associated with the small mobility workflow, as well as all the results overall. It allows us to look at impact on the e-commerce travels, on mobility, on congestion, on energy, on travel time, on fuel savings, on greenhouse gases and cost, on many, many, many different parameters, you know, as well as trying to aggregate them into a single one so we can compare all the different scenarios. So with that in mind, thank-you for your attention, and I believe we can move to the Q&A.

David Anderson:
Next slide:
Yes, we can. Thank-you very much, Aymeric. We do have about 13 minutes remaining for Q&A. I have received some questions in the Q&A box. I'll remind you to submit your questions in the Q&A panel in the Webex. First question, I'm not exactly sure how to interpret this first question, so we'll do it in two ways. The question is, are the simulation models online or offline? And so I guess one way I would interpret that as, are the models that we built available for download for others to use? This was asked early on in the presentation. So some of it might have been answered, you know, as you described the workflow. Another interpretation might be, are these models run on local computing resources or are they run on cloud computing resources?

Aymeric Rousseau:
Yeah, and the answer to both question is it depends. So right now, some models are run on a single desktop; some others need high-performance computing. And the one only that performance computing, some of them are on the cloud, and some of them are not. And then in terms of availability, also the answer is it depends. When we go to some of the models are freely open and available, some of them have to be licensed. So one thing we're trying to do, you know, in the follow-up smart mobility consumption, is really to be a lot more consistent so we can deploy not just some tools but all the tools. And within the workflow to outside organizations, and so that's going to be one of the major efforts. The effort in the past three years was really to connect all the dots and then start to exercise then improve the linkages. And now that we are at the point a lot of the efforts will be developing. Some, I would say, you know, it's not commercial grade. But at least some workflow that are available to our organization from end to end. So you can also exercise the workflow.

David Anderson:
Alright, thanks. We do have a number of questions coming in, so I'll remind everyone, if we don't get to your question feel free to submit it to the email address that is shown on your screen there. Another question is, does the mobility models consider weather data since the type of mobility transport could depend on weather, and it could also be seasonal.

Aymeric Rousseau:
Yeah, so that that is a very good point. You know, if we're looking at riding a bike, you know, in Chicago, you have to make sure that, you know, we don't model February, right? I mean, there may be some people out there, but definitely not that many. So in terms of both models, and I'll -- so by the way, Joshua Olds and Zach Needle are on the panel, as well. Josh is leading the implementation of Polaris for Argonne, and Zach has been leading the implementation in Beam for Lawrence Berkeley. Please chip in. But Polaris, I know for a fact, has the ability to integrate snow, for example, where it will significantly slow down vehicles on the highway and on in other areas, as well as affect some of the modes. So that is something we can do. I would say that is not something we've included so far in our scenarios, for mobility. But, you know, when we're looking at new modes coming up, whether it's scooter or e-bikes or so on and so forth, it becomes a little bit more important and that's a good point.

David Anderson:
Great, thanks. We have a question; it might be meant for me, but I will defer to the folks who have developed our modeling capability: What are some of the ideas or plans that we have for putting these tools in the hands of users like MPOs, which I alluded to initially? They're currently -- I would characterize them as research tools, but again, ultimately we want to deploy them, yeah.

Aymeric Rousseau:
I'll speak in general and maybe more particularly, on the workflow at Argonne, since that's the one Josh and I are involved in. So in general the first point is that both workflows have been recently expanded to additional metropolitan area. I mentioned that Polaris was really focused on Chicago and Beam on San Francisco. Both models have been expanded to new areas. Two of them in common, Detroit metro area and Austin metro area, that both tools have been expanded to, and Polaris has also been expanded to Atlanta, Knoxville, Chattanooga metro area. All of these have been done in very close collaboration with training agencies, right? Local universities and researchers are involved, planning agencies are involved, and the main reason is we want to make sure that when the models are available, which you know they will be literally within a couple of weeks for these new cities, the planning agencies have them, have access to them and understand how they've been done. In terms of deployment, you know, there is obviously different needs when deploying tools to developers vs. users. So right now, as David mentioned, most of our tools are really developer tools, right? You need to get your hands dirty, do a little bit of coding. What we want to do in the near future is actually develop graphical horizontal effects. And whether it's cloud-based, whether it's desktop-based, we want to make sure we develop graphical user interface so that we don't show, you know, an infinite number of scenarios with an infinite number of parameters, to users, but we show very specific scenarios and very specific users. So that will be done. We have a very large group of concession that will be very important to us, and that goes from, that includes planning agencies, that includes a research organization with universities, it includes mobility companies, as well as private companies will be providing us with requirements as to what parameter, what scenario they're most interested in, so we can develop user interface, which I think would be very key to deploying all the tools to a much larger community.

David Anderson:
Alright, sorry, I'm trying to triage here. There's a question that relates to transit, and I don't think you have the slide in the deck, but we've done some work here. And the question is, do we have examples of the effect of transit in some of our scenarios? So maybe you can verbally respond to what we've done and what we intend to do.

Aymeric Rousseau:
Yes, very good question, and transit is a hot topic right now. But one of the scenarios we run in Chicago is, what happens -- and that was demand with being pre-covered as part of the small mobility -- what happens when in the city, when there is no transit? And one way to look at it or to say it's Armageddon, right? Downtown Chicago gets gridlocked, you got traffic congestion that goes through the roof. My memory is correct; it's about 50 percent increase. So it's literally, transit is critical core and center to transportation system. And you know, it's a lot more efficient to move people universe than to move in individual vehicles. And we've significantly expanded that study with COVID on a parallel effort to look at the, you know, the impact of, you know, when and how and where people go back to transit in different COVID scenarios. But we've done some significant amounts of studies, specifically related to transit, and we understand this is becoming a much bigger issue. And so we will be spending a lot more time and focus on the transit in the future.

Additional participant:
Aymeric, this is Josh, and just to build on the answer directed to the question that was asked: Some of the -- all of the scenarios that were considered also looked at the effect of, for example, increased vehicle sharing, increased private AV ownership, and how that affects the transit system in each case. So multimodal routing and vehicle use was definitely considered across all of the scenarios, and I think that will be expanded upon in some of the later webinars. It just at a high level didn't show up in this one.

Aymeric Rousseau:
Yeah, thanks, Josh.

David Anderson:
Yeah, thank-you, Josh. Wow, lots of questions coming in. So we have a question on, how are we considering the value of time for automated vehicles, and how do we apply this value to the land use models in terms of residential choice? I don't know if you can start with that.

Aymeric Rousseau:
I will let Josh take that one.

Additional participant:
Yes, definitely, the value of time and the fact that having vehicle automation or different levels of vehicle automation, and even in different scenarios, whether it's congested highway driving or open road driving, we consider all of that in the scenarios that are run, with the AV having different impacts on the OT in those different scenarios. Not a lot of empirical data on that. There are some studies been done, some studies being done currently, on how people behave in different types of vehicles. But there are some assumptions we make around relieving some of the pressure or disutility of driving when you have automation in the vehicle, that we consider. And that flows through to the land use model in terms of the generalized travel times and costs that are passed from Polaris into Polaris or Beam into UrbanSim. So both in the model skims and in the link cost. So we send an actual travel time and then a generalized travel time, which is affected by the amount of people that are traveling in AVs at the lower time cost.

David Anderson:
Thanks, Josh. Hopefully that answered that question. Another question we have, which I have now lost, um ... is, do we have the capability to create an idealized vehicle with powertrains and controls? Can we automatically create hardware and operating strategy requirements for an ideal vehicle for a given environment, that is, driving location and application such as delivery or private vehicle, etc.? Can you speak to if our workflow is suitable for that, or if there are other tools in our arsenal that would be appropriate?

Aymeric Rousseau:
That is a good question. One of the tools that would be used for this is Autonomie. So vehicle energy consumption model, and within Autonomie what we do is we build vehicles from the ground up to match specific vehicle technical specification. So think about, you know, your acceleration, your credibility, your passing, but also your range. So when we simulate. And to be able to design vehicle for specific applications, right, you need to understand what their usage is. So tools like Polaris allow us to understand vehicle usage today but also tomorrow. Once we have, let's say, many connected automated vehicles on the roads, we may not need to have such a high performance for the l260, right? We may not need to have eight-speed automatic transmission; maybe six or four is enough, because you have a much smoother environment. And maybe because also you most likely are electrified. So based on the requirements that we see out of Polaris in terms of range and distance, acceleration, deceleration, we can actually design new vehicles based on the new set of vehicle technical specification. So instead of having, for example, 0 to 60 at 8 seconds, maybe we can do 10 or 11 seconds and do that for all the scenarios. In terms of freight, it's the same thing, right? We have many different freight options, and the route length is actually changing quite a lot right now. So the average distance for freight, especially, you know, the daily ones, is getting smaller and smaller because of the way the supply chain is being redesigned. So we also can take that into account to design vehicles and powertrain that meet tomorrow's demands, not just today.

David Anderson:
Great; thank-you, Aymeric. I am showing 4 o'clock Eastern time. We have lots of additional great questions in the Q&A box; unfortunately, we won't have time to get to them. I do invite you to submit any questions to the EEMS email address that is shown on your screen. I do want to remind everybody, if we could go to the next slide ...

Next slide:
... that we have published a set of six capstone reports that provide all of the details with the work that we've done. Five reports focus on the five pillars of research that I talked about earlier in my introduction, and then the sixth report goes into details on the modeling workflow, its development, its implementation and results. The link shown there on the screen, which is also repeated here, will allow you to download these reports and also allow you to register for upcoming webinars.

Next slide:
Our next webinar will be held on October 8. It is focused on the MEP metric, introducing a novel metric to quantify the impact of smart mobility. That's the mobility energy productivity metric. And again, every two weeks, webinar 3 and webinar 4, focused on applying the smart workflow and understanding connected and automated vehicles. That concludes our webinar for today. Thank-you all for joining us. Hopefully you found it informative. Thank-you, Aymeric, for presenting, and hopefully we will see you all again in two weeks. Stay safe and be well.