All right it's just a few minutes after the top of the hour so we will go ahead and get started. Thank you everyone for joining us as we get started. I will remind you that this WebEx call is being recorded and may be posted on VTO'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 choose to participate by phone. If you speak during the call or use a video connection, you are presumed to consent to recording and use of your voice or image.

Again, good afternoon! I’m happy to welcome you all to our fifth webinar in a series in which we summarize the work and the results from DOE's Smart Mobility Lab Consortium. My name is David Anderson, I’m the program manager for Energy Efficient Mobility Systems or EEMS. EEMs is a program of the Vehicle Technologies Office which is part of DOE’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 time and the affordability of transportation. We've been having these webinars every two weeks and we've covered the development of our smart mobility modeling workflow and the mobility energy productivity metric we've covered results from the two implementations of the workflow and specific research on connected and automated vehicles. Now today our focus is on moving people in the US.

Vehicles moving passengers travel 3 trillion miles every year and this passenger volume shares our roadways with 11 billion tons of freight. Now freight will be the focus of our next webinar in two weeks, but today it's people that we're focused on our two speakers today are Dr. Anna Spurlock from Lawrence Berkeley National Lab. Anna led the mobility decision science pillar of research under smart mobility 1.0 and Dr. Josh Auld from Argonne National Lab. Josh was a key architect behind the smart mobility modeling workflow but before I turn it over to Anna and Josh I’ll again give you some context.

As I mentioned at the beginning EEMS is one of the research programs within the Vehicle Technologies Office or VTO. Now many of you by now are aware of the R&D that VTO has done successfully for a long time on vehicle powertrains and component technologies themes and smart mobility extend VTO’s scope up to the transportation systems level. We look at technologies not only at the vehicle and component level but also how vehicles interact with each other and with infrastructure at the small network and corridor level. We can then evaluate the impact that new solutions have on overall traffic flow and energy consumption and finally we bring people into the equation and we work to understand the future of mobility in the context of an entire urban or metropolitan area. Integrating passengers in freight including on-road vehicles in transit micro-mobility and pedestrians and biking. The entire mobility system.

So that's the purview of the EEMS program now to make sure we execute on that charge we've taken a multi-disciplinary consortium approach to transportation research. We convened five of our leading national labs including dozens of our most skilled transportation scientists and engineers to build the smart mobility laboratory consortium. That's a multi-year, multi-laboratory collaborative dedicated to further understanding the energy implications and opportunities of advanced mobility solutions. The first phase of our consortium was divided among five pillars of research shown here they are connected and automated vehicles mobility decision science multimodal freight urban science and advanced fueling infrastructure the research results that Anna and Josh will describe today largely come from the mobility decision science focus area, but since the consortium is multi-disciplinary in cross-cutting these focus areas were very much interrelated the research in any one area relied on activities from the other areas.

So that's why I think it's more important to look at the consortium and mobility research in general like this as a complex system of systems we can't really study any one area in isolation without considering the impacts that related areas have on it. Which is what we set out to do through the consortium, it's a bit of a happy accident that the traveler behavior circle is the brightly colored circle in the middle of this diagram since it comes down to people in their decision making which ultimately drives most mobility outcomes.

Now I keep using the term mobility so I want you to remind you what we mean we've defined 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 this definition is important because it emphasizes our focus on moving people. That's the topic of today's webinar. We're working to connect people to the things they need better and to do it affordably and efficiently to give people access to opportunities like jobs in education, health care and recreation and shopping and retail mobility is fundamental to our way of life. Our presentation today is divided into two parts. First Anna Spurlock will talk about some of the empirical data analysis in the whole traveler behavioral study results that were created under smart mobility. She'll describe findings on topics like how do people decide which travel mode to take what factors influence technology adoption what lifestyle trajectories lead to vehicle dependence. What are the barriers to EV penetration what impact does ride hailing have on vehicle ownership and energy use and finally how have consumers e-commerce decisions impacted mobility.

After Anna concludes, Josh Auld will discuss some of the modeling and simulation results generated from the smart mobility modeling workflow relevant to passenger movement. The smart mobility modeling workflow is an end-to-end, multi-fidelity software tool chain built around agent-based transportation system models to capture the complex interactions among mobility decision making. Technology implementation different mobility service models different modes land use and EV charging infrastructure. It integrates freight and passenger movement and allows researchers to evaluate the energy mobility and portability outcomes of potential future transportation scenarios. We can evaluate these outcomes in terms of vehicle miles traveled travel time energy costs greenhouse gas emissions and other metrics.

Now within the context of moving people, we've asked questions like in a future in which vehicles are connected and automated, will people buy them? How will they use them? Will they change our travel behavior as new mobility services reach maturity? What does that mean for how people travel and for how much people travel? What does it mean for the number of cars on the road?

In addition to new services what about the new modes being introduced, will micro mobility become a viable transportation option? And if so, will it affect other modes such as transit or will it just serve as a niche role?

And what is the role of transit we believe public transit is the backbone of any urban transportation system, so how do we make it more accessible and convenient for people and given the pandemic's impact on transit what is the future of public transportation?

So, with that background I’ll now turn it over to Anna Spurlock from Lawrence Berkeley National Lab to discuss the empirical and data empirical data analysis and research findings from the mobility decision science area. Once she's done she'll turn it over to Josh Auld to cover from Argonne National Lab to discuss simulation mixed results at the end we'll have a few minutes for Q&A and I encourage you to type your questions into the Q&A box on your screen, not the chat box but the Q&A box and we will get to as many of your questions as time allows. At this point I will turn it over to Dr. Anna Spurlock.

Great thank you very much! So, thanks everybody for joining as David mentioned I’m going to talk about the work done in most of what I’m putting on is work done in the mobility decision science pillar from smart mobility and to give a little bit of an overview the idea of that pillar was to examine the underlying sort of factors. The preferences identity factors personality life cycle contextual and geographic factors that contributed to people's transportation behaviors and in particular to understand the critical drivers of or critical barriers to adoption use of emerging transportation technologies and services and then to think about what that those barriers and drivers mean for the implications of these technologies and services for the system in the mobility decision. We tackled a range of topics it was a wide-ranging set of work and so some of the work done can be categorized into these topic areas. So, life context and what that means for transportation behaviors, psychological and identity factors with regards to transportation preferences. E-commerce use and shopping travel and the impact of hailing on a transportation system and in terms of methodological approaches. We used a variety of approaches in this pillar, so there was the whole traveler transportation behavior study which David mentioned which was a data collection, a survey data collection effort in the San Francisco Bay area. So there was an original data collected from that study and then a variety of analyses that touched on all of these topics done using those data in addition there was work done using the agent-based transportation system simulation models that we have including the full workflow that David mentioned and then there was also standalone stand-alone analyses on other types of data sets touching on a couple of these topics which I’ll also talk about briefly now.

So, as I say it. was a there were a lot of things that we covered in this research and so we won't be able to talk about all of it today but I will try to bring forth some of the more interesting salient points we had. So, one thing that we did look at was you know life cycle and conceptual factors for people's interest in and adoption of different transportation technologies and services. So, when we just think about the kind of main thread that carries us through our life cycles as we you know kind of as we age one of the things we were looking at was how people in different sort of generational categories reflected in terms of their interest in an adoption of a variety of different technologies and some findings were relatively expected, like consistent with other research for example younger generations. Here the millennials tended to drive adoption of ride hailing in our survey but we did have some interesting insights that tended to contribute to question marks that existed or gaps in the literature and one thing we found for example was that baby boomers, which are the older of the generations that we looked at even more so the generation X had a higher interest in and rate of adoption of advanced transportation vehicle technologies like electrified vehicles. So this was something we found interesting and why it matters is when we think about how people use transportation alternatives over the course of their lives we see that as people age they tend to move towards more and more vehicle dependence. So, in our survey data what we showed is that as people reached about 33 in terms of age they tended to kind of reach a pretty high rate of regular reporting. Regularly driving their personal vehicle for their main travel and in our survey data that was about 70 reporting that by about that age and then it stayed kind of consistently at that level or even slightly higher thereafter.

So what's interesting about thinking about preferences for new vehicle technologies is if we think about the fact that there are factors driving this shift towards more private vehicle dependence with age, understanding whether people as they get older maintain or have a high interest in different types of vehicles that might have different levels of sustainability or impacts on the rest of the system or environment can mean that there are interesting implications for or important implications. For that  kind of system level energy or other factors when we think also about people progressing through their life cycles. One of the things we think about is you know, basically differences across people in how this happens. So if we take all of the participants in the survey who we used in this particular analysis that I’m going to talk about and we kind of dump them in a bucket and look at how they progress through key life cycle events or phases, such as being in school, being employed ,living with a partner, having children… this is what you get sort of on average in terms of their average progression through these phases. But what we did in our analysis is we looked at trying to understand whether there were you know key different patterns in which people progress through these phases.

So, we use a clustering analysis to look at to basically characterize five different archetypal patterns that people fall into when they progress through these phases. So, to understand this if you think about the first three here those single couples and have it all which we named just because of the nature of their patterns, one of the things you observe these are the three most sort of prominent in terms of the% of the sample that fell into these patterns. You see that these streaks progress through their career development a relatively similar pattern. They all finish school and start working in a similar rate and at a similar timing in their life, but they all differ significantly with respect to their family formation patterns. So, the singles tended to wait until later in life or choose not to live with a partner or have children. Couples on the other hand tended on the other hand, tended to couple up relatively quickly similar in timing wise to their sort of career formation, but tended to wait to have children or chose not to have children, and then have it all. On the other hand, tended to couple up again relatively early right around the time they were starting to work full time but very quickly thereafter had children and most almost all of them had children. There were two other smaller cohorts. so what we refer to as the late bloomers which tended to delay most of these life cycle stages relative to some of the other cohorts and then the family first which tended to have family formation precede sort of schooling or career.

So what some of the interesting we learned a variety of interesting things from analyzing these cohorts but I’ll just give you one set of interesting take away. But if have it all cohort was in particular interest and one of the reasons that that this is the case is because one of the things we found was when you look at the patterns of the mode, use of these cohorts they have it all cohorts you see a relatively steep increase in the rate at which they achieve or I don't know achieve. But the rate at which they kind of move shift towards this more significant vehicle dependence in their driving styles and in particular what we found is that this cohort more than any of the others statistically significantly increase the rate at which they were reported regularly driving at every one of their lifecycle transitions. So stop you know finishing school, becoming employed full-time, living with a partner and having children, all resulted in a statistically significant increase in the rates in which we were reporting regularly driving and the implication of this is that this cohort you can see reached the highest level of vehicle dependence the earliest of all of the cohorts and stayed there.

So when we think about the underlying needs that that people have for transportation alternatives that are kind of dependent on the different forces that are acting in their lives based on their choices of how they progress through their lives and what responsibilities they have and the things they're trying to accomplish. You can see that it can have a long-term implications for the kind of impact of their choices with respect to energy for example one of the other things we looked at was gender with respect to these different cohorts and just to give an example of an interesting finding here so the women in the habitual cohort tended to be even more vehicle dependent than the men in the same cohort. In particular this related to the timing of them having children. So when having children are preparing to have children and women in this cohort even more so than the men intended to engage in more family member transportation responsibilities, they use public transit even less they move to public transit poor areas even more they tended to drive even more and live households with more vehicles and live in households with bigger vehicles. So you know for us this was an interesting window into thinking about that idea of the critical drivers of our barriers to certain types of transportation choices and availability of certain alternatives, not just physically in terms of availability in terms of how close you are to that to those alternatives, but also availability with respect to whether they could fit into the type of lifestyle that one is living in.

Keeping on the theme of gender but shifting a little bit to a new topic one of the other things we looked at was the barriers to electric vehicle adoption and so one thing we observed in our data which has been shown by a variety of other studies that have looked at PEV adoption or plug-in electric facial adoption or interest is that there was a gender gap. We had about 15%age points size of difference in terms of the rate at which women in the survey responded that they were interested in adopting an electric vehicle compared to men. So we wanted to dig a little bit into what the nature of this gap was what was driving it so we hypothesized a variety of underlying factors. One risk preferences of a variety of types personality characteristics willingness or ability to pay transportation needs such as habits or the responsibilities in the household and environmental attitudes, and we looked at this using what's called a mediating analysis or mediation analysis and what we found is that of all of those factors we entered in that entered into the analysis they explained about 35 collectively of the gender gap. So there's definitely more you know contributing to this that we don't necessarily understand yet, but in terms of some of the interesting takeaways from our analysis we found basically and we can frame it this way. That if women have the same income on average as men and in our sample they tended to have on average lower income than men then the gender gap would be about 10% smaller and women had the same responsibilities for transporting family members that they need for cargo space and vehicle capacity and the same air and running responsibilities as men which they tended to have higher need for those things than men in our sample. The gender gap would be 15 smaller, so this you know points to some of the interesting the interesting underlying factors that might be contributing to people's willingness to take up different alternatives. If they're not kind of geared towards the needs or requirements of a given subpopulation.

So, I’m going to shift gears again a little bit here and talk a little bit about ride hailing. In the whole traveler study we did look at we did an analysis looking at ride hailing we did in particular the impact of ride hailing on public transit and we did show that ride hailing can be both a compliment and a substitute for public transit and it depends on the relative's price and it depends on the proximity to a public transit or kind of a degree to which is likely to impact preceding behaviors. But I’m not going to, I don't have a slide on that particular result. I’m going to shift to a couple of other analyses that were done with other data sources under the mobility decision science pillar. So one looked at the entrance of ride hailing services to urban centers across the US and basically found that that basically that for urban areas with lower per capita vehicle ownership and with higher rates of economic growth ride-hailing entering the market increased the numbers of vehicle registrations by about 0.7 on average.

So there was variation across different urban areas. So you know that we weren't able to necessarily nail down the exact reasons for this but you know it begs the question of in these areas was it possible that people who were interested in driving for this service actually tended to acquire a vehicle that they wouldn't have otherwise and that could have contributed to the increase in people registrations? So this speaks to the fact that while you know ride hailing may have impact the system through the behaviors of the riders like the whole traveler result I mentioned they can also impact the system through the behaviors of the drivers of the service another similar themed results came from analysis of data from a ride hailing service in Austin Texas called Ride Austin and what was found in this analysis is that mainly contributing to the fact that ride hailing drivers commuted relatively long distances and also due to the repositioning behavior of those drivers the sort of deadheading behavior between picking up riders ride hailing resulted in an increase in system-wide energy consumption by 41 to 90%. And the variation between these depended on the assumptions used in the analysis on the rate of cooling or the modal shift that resulted from ride hailing so those two results you know speak to the fact that there are a lot of different actors contributing to how these different alternatives function in the system and they can all have implications for the outcome.

Finally touching on a topic that is a shifting quickly shifting topic in our society right now looking at e-commerce in the whole traveler's study what we were looking at is the degree to which e-commerce with delivery substituted for or resulted in additional deliveries without reductions in shopping trips. Now one caveat being that this is all pre-covered e-commerce delivery context, so I imagine things are much different now but at that time what we looked at was we categorized different types of shopping. So shopping for clothes groceries household items or prepared meals and you know this is basically just showing the relative sort of frequency of these types of shopping trips reported in our sample and the mode that was used for those shopping trips. So particularly what we found is that groceries not surprisingly were the most frequently purchased but the least frequently done so through a delivery. On the other hand clothing were least frequently purchased but proportionately more so via delivery. And when we look at the pattern of substitution what we found is that delivery to households both substituted for and was in addition to shopping trips and but that what we found in aggregate is that a typical in a typical week given delivery was about 1.7 times as likely to substitute for a shopping trip than not and of those that did substitute a given delivery was about 300 more likely to substitute for vehicle trips than a non-vehicle trip. So in aggregate the substitution result of delivery did kind of win out against it being additional travel or activity in the system but that depended on characteristics of the household.

So I’m going to shift to passing over the presentation to Josh to talk then about some more of the system impact results of some of these topics.

Okay. thank you Anna! So the rest of this presentation will really focus on behavioral findings that come through the application of the simulation models that have been developed as a part of the smart mobility program that's the workflows that are centered around the Polaris and beam activity-based travel demand models.

So, the behavioral research that has been done is largely in service of improving and extending the smart mobility workflow. As I mentioned, this includes taking findings from survey data other simulation study analysis and other sources and developing models of a wide range of demand and behavior-focused aspects of individual and groups. The goal being to take these models and extend the existing smart mobility workflow to both capture these behaviors as well as be able to explore these behaviors and how they interact with other behaviors. So not doing things in in a silo but as a system of systems as was mentioned in the introduction. So, a lot of focus has been put in the mobility and decision science pillar in terms of developing new models of mode and timing choice value of time and time use telecommute behavior. Vehicle sharing behaviors, e-commerce as Anna just mentioned and vehicle technology choices and these have all been incorporated in the smart workflow and the impacts have been evaluated through scenario analysis as well as through evaluation of the sensitivity of the overall model outcomes to both key parameters for each model and parameters across models. Important to note here and there is a reference slide later in the presentation that many of these findings were not only used in service of improving the system modeling workflow, but also were generally researching themselves and published in in multiple peer-reviewed journals. Just some highlights so won't go into every behavioral study that's been done and incorporated into the simulation models but some of the highlights here include work that has been done on exploring the value of travel time savings and time use that have been explored from a perspective of different modes and setting up for the what-if analysis of time use and time shift in new modes.

So a few papers by Kruger at all when in amid all recently published have really been exploring how we incorporate different data from multiple sources. Whether it's household travel surveys time use surveys onboard transit surveys or more speculative surveys to develop an integrated choice and latent variable framework to quantify not only the value of travel time but how the travel time shifts under the multitasking and time use opportunities posed by different modes. So some of the key findings we get out of this research are that, one there are significant variation in terms of what we find for the value of travel time when we look at different sources, but generally across sources we see that the ability to multitask encourages use of non-drive modes as well as reduces the value of travel time experienced on those modes and that plays into the idea of future automated vehicles serving as a multi-tasking and new time use space that could potentially lower the cost of travel in such vehicles. There's also been substantial work on improving baseline mode and timing choice models some of that published in Kosrani at all 2018.

So we have a mode model developed for Chicago from the Chicago household travel survey that uses the nested logic model to look at the choice between different modal options in a nested framework and we find here that the value of in vehicle time varies substantially by mode activity purpose expected. Travel time and congestion and travel time variability and so on. A lot of these different system level factors as well as individual factors that help to explain how in-vehicle travel time is perceived by different travelers in different contexts. We also looked at modeling a joint start time and duration model to say what time of day people choose to travel for, and how long they engage in activities at the end of that travel and this key finding here was that this depends strongly on the activity purpose as well as the expected travel time and travel time variability in reaching that destination.

So accessibility and access to high quality activity opportunities strongly influence when and how often people tend to travel there's also been work at looking at how AVs as we focus on automated vehicle technologies how those would impact household behavior. One of the initial studies being done on sharing in households to see how households could optimally use an AV that has the ability to reposition itself to minimize cost for daily travel in terms of the number of vehicles required and operational costs, parking costs, time costs and so on. To say how can households shift both their activity patterns their activity start time and durations to get the most use out of a single vehicle and this was implemented in external optimizer and many studies including all of our 2019 deriving findings out of that finally one other highlight of the behavioral research that went into the simulation frameworks is looking at e-commerce and this will be presented in much more detail in the next upcoming webinar.

But some work that Stinson at all had done in conjunction with the whole traveler team looking at how households participate in e-commerce as Anna mentioned in the last slide and looking at how this displaces physical shopping trips. So, exploring that through a system framework to see what is the overall impact and what we find is that the desire of our households replacing physical shopping with deliveries is dependent strongly on income. The number of vehicles that household owns the location where they live and then their accessibility to both retail and transit travel opportunities and that the overall outcomes from replacing those shopping trips have varying impacts in terms of overall system travel and energy efficiency depending on where they live and how frequently they shop. So that was a lot of the background work that went into improving the workflows, the modeling workflows developed as part of smart so that had been presented in previous of the webinars in the webinar series we then applied these workflows as has been presented previously to a variety of potential. Almost corner case feature scenarios looking at different levels of sharing and automation and as well as other technologies and how those interact. So we have a near-term case of high sharing but partial automation, So this is a slightly more automated but much more expanded version of the existing TNC fleets. There's a high sharing high automated version which is basically that same scenario, but then we automate the vehicles and have a lot of advanced technology in the vehicles where they can operate fully autonomously and as well as having expanded e-commerce opportunities. And then there's a low sharing version of that high automation scenario where we have the same advanced vehicles but now they're privately owned by households.

Through this simulation and analysis we find that these individual traveler behavior changes drive strongly the ability outcomes. So just looking at some of the behavioral models and how that filters through these simulations we should see that as we look at these different scenarios. Mode share changes substantially driven by a number of different factors, so we see that transit use grows from about six% all the way up to twelve% when we have both high sharing and high automation as households dispose of those vehicles as they come to rely on the automated vehicle service. But then can actually take more transit trips since they do not have the vehicles for land anymore when the transit is the more efficient option. Conversely, we find that private AV does encourage additional shared private vehicle trips even in the case where household vehicles are disposed.

Now we're back to almost if you sum up the all of the privately owned vehicle trips, it'll be about seventy% versus the seventy five. So eight% of the baseline and we do see that urban households shift to transit while suburban households shift larger the tnt although there is some mixing driven by this vehicle disposal and a lot of the impact of the shared vehicles comes from this assumed lower value of travel time savings this is the monetary cost the traveler would pay to avoid spending additional hour in travel. But one thing is that this value of travel time savings differs greatly by mode income location trip purpose and so on and what we find when we run a simple scenario analysis in a Polaris workflow that if we reduced VOTT by 50 across the board it doesn't mean people travel 50 more it means some households for example in the core suburban areas we can see you know these outlying areas around city of Chicago have about a 52% increase in their overall travel as they're less deterred by spending more time in the car. They travel further to more high quality activity opportunities whereas people in the downtown area only travel about a third more so the average increase of about five miles per capita because they're already basically surrounded by a large amount of high quality heterogeneous for different activity purposes activity location. So there's no real need to travel further even though traveling further would not be more costly for them to reduce VOTT. So this is highly geographic and context dependent in terms of what impact this VOTT shift would have playing into that or building on that we see that the automated vehicles have this made their effect on travel behavior because of this assumes ability of people to refocus from the driving path to other productive time uses tasks and even if it's not another non-productive time use it's still less burdensome to not have to drive and focus on the road which lowers the cost and as we saw in the previous slide means people travel further for their activity opportunities if it's beneficial.

So what we see when we run through the high automated privately owned AV scenario that the average household travels about 82 more miles on average, 57 of those are person miles. So the actual people traveling further and then the rest from the vehicle repositioning and then if we compare those households with the households that don't own AVs even in the same scenario. We see this effect pretty clearly, so this 82 increase in households that own an AV is largely driven by increased discretionary activity trip making. So as we mentioned going to a better restaurant a more advantageous shopping location you know traveling to see more distant friends we have an average three to six mile longer per trip travel time an increase of about 30% from baseline for largely discretionary trip activities and so we see that the average person with an AV spends about 30 minutes more per day and travel due to that lower travel time cost and more opportunity for engaging in other activity types in a related finding through the beam workflow coupled with the urban sim model. We see that the lower the lower subjective travel cost of the lower VOTT also would affect the demand for land outside of the urban core so all the scenarios where we're reducing VOTT have the effect of increasing accessibility through the reduced generalized travel time costs and this would lead to more demand to shift residences and businesses further from the urban core which was most prevalent as shown in the chart right in the high sharing high automation scenario where basically people due to the lower travel time cost can reach further jobs in less dense areas. Which has the effect of increasing land values further from CBD and inducing urban sprawl now to counteract.

That one other key finding of our behavioral research is that transit is critical to mobility so one way to demonstrate this is we ran a scenario in the Chicago model without transit and we find that without transit the energies and congestion increase in the city of Chicago. It's an unacceptable level so changes in link level speed throughout the Chicago network shown on the map at left here and we see that in the city of Chicago without transit there's about a 29 increase in energy use, a 37% increase in miles traveled and a 52 increase in hours traveled to a substantial increase in congestion throughout the region. However this is not it doesn't have to be the case so transit and right hail, so even in in a world of increased ride hail and automated ride sharing deployment we see that transit and right hail can be actually complementary and work together to provide key mobility. Especially in denser areas where we see that transit still is a key mobility provider in the city and along inter-corridors whereas TNC serves the less dense suburban areas so if we look at transit ridership and TNC ridership across our moderately automated and high automated shared vehicle scenarios we see that transit mode grows fairly substantially in the areas of high quality existing transit and along the main sort of hub and spoke system in Chicago even as TNC use also grows in many of those in many of those areas where these form almost mirror image of each other where transit growing in this area and TNC being used extensively everywhere else.

So that then this happens even as TNCs increases in these areas as well so some in this case indirect complementarity we also explored a direct implementation between transit and TNCs so we found that integration of these systems can produce benefits beyond what would just be seen in in the previous simulation study in this case we applied it to our small-scale Bloomington Illinois sort of a moderately sized city in Illinois about a hundred thousand people with a fairly low density low utilized transit network about 1.3% of people taking trips by transit in a baseline and it's all bus network what we found in this case is we could have a system where we integrate first mile last mile service from TNC with the transit fares and increased transit ridership by about 11% through only adding about four thousand extra VMT miles. So about four thousand extra miles on into the system from this first mile last mile access would reduce total VMT by about three thirty three thousand miles so about a 1.4% reduction total VMT 1.1% reduction in fuel and 11% increase in transit mode shift with this moderate integration. So we found that this could be an effective pilot effective strategy that could be put in place to boost transit ridership in a low utilization scenario conversely again I’m finding through the beam workflow we find that reducing ride hail usage does also overall affect transit ridership and total energy.

So when we reduce ride hail we result in lower use of transit and a higher use of personal vehicles leading to in a high reduction case where we have about a 75 reduction in the pool of travelers considering ride hail. You have an overall increase of 6% in terms of energy use so we do find although that in some instances ride hail trips can produce more VMT than car trips the loss of this ride hail to transit mode offsets this that potential downside meaning that overall our system is worse off when many fewer people are considering ride hail and then just to build on the previous the work that Anna had mentioned in the first portion of the presentation. We also looked at this this e-commerce on a household level and how that would filter in through system findings and we found that when we have more e-commerce deliveries overall you can actually in many instances reduce system miles traveled and energy use and that's due to replacing shopping trips with more delivery trips where whereas even though the delivery trip is generally implemented with a less efficient vehicle, usually some type of medium duty box truck adding additional stop on a on an efficiently planned delivery tour where there tends to be 120-140 stops only would add about a less than half a mile.

You know marginal increase in miles travel on average to that delivery tour whereas the average shopping trip in simulation findings and survey analysis looking across the country tends to be about seven to eight miles one way. So replacing this if it was a one-to-one replacement with a shopping trip versus an e-commerce delivery now there's a clear gain. It's not always one-to-one but there's clear gains in terms of overall VMT and energy use and again we'll go more into detail on these results in the next in talk in the webinar series. So just some high-level takeaways before we move into the Q&A portion. The idea behind all of this work was really on trying to understand the motivations and constraints of facing individual travelers to lead to more realistic scenarios when we do our scenario analysis of future technology deployment and future let's say urban mobility design. So we understand that people make the choices they make for a reason and the idea is to try to capture those through the workflow that we use for future what if scenario planning analysis.

So in the smart in the initial set of scenarios we did look at a limited set of edge cases some of which assumed relatively high willingness to use certain modes or technologies such as shared ride hail and made fixed assumptions in terms of shifts and value of travel time and so on. But the results from the MBS demonstrate that the extent to which people might face constraints given their context or life choices may limit the actual ability to adopt these different technologies that are specified in the scenario design for some of these alternatives. So it's going to be good to continue looking into these individual you know unique heterogeneous reasons that people make or don't make choices to understand how those would filter into the scenario analysis. So understanding why these constraints can inform which scenarios are realistic and why to do simulation under analysis so this this the simulation findings really underscore why this deeper understanding matters when we're trying to look at system wide outcomes for planning and really understanding what the deployment of these different technologies. What the impacts will have and better managing those so just a couple quick references and won't read through these but they're available for those who want to look and download the slides this is you know a sampling of the work that's been done under the VMS or the MDS pillar as a part of smart mobility and with that I’ll say thank you and we can control.

All right thank you Josh and thank you Anna for those presentations. We have just about 12 minutes for Q&A and so we'll go ahead and get started with that. Again, I’ll remind everyone to enter your questions in the Q&A box on your on your screen the first question we have is for Anna. The question says: “did your study take into account impacts on wealth for the different generations?” The questioner would be curious to know how things like the recession that was experienced a little over a decade ago how that impacted for example the millennials ability to earn wealth which may impact adoption of more expensive technologies. So Anna do you have a response to that? Yes that's a good question! So when we looked at the results that I presented with respect to those generations we did do an analysis that controlled for we didn't control for wealth but did we did control for household income. I would say that the main that that that factor asked about is definitely going to be important from the perspective of actual adoption of these technologies. The analysis I did present on was more about people saying they were interested in adopting the technology, So that's a little bit of a different angle on it but I would agree with the questions premise that that wealth and the ability to pay for these things is going to have a big impact on the actual potential kind of market penetration and roll out of these technologies.

Great thank you. I have another two questions that are very similar. So I’ll go ahead and ask them both they're really for both of you but maybe we'll start with Josh and then you know weigh in. So the two questions are “how are the results from the empirical data used to make better models? Do you have an example and so the empirical data that Anna discussed how's that used to make better models such as those that Josh discussed?” The other similar question here is “have any of the insights that Anna presented informed any of the specifications or the scenarios for the simulation results that Josh presented?” So I’ll start with Josh can you speak to how the data efforts that Anna described and feed into or inform the model work modeling like that useful? Sure yes, there's a pretty direct connection there you know one of the examples like you know that we touched on in the presentation is around this issue of e-commerce. So the data we used in estimating that model was directly from any and even some of the findings that used to inform the model structure was directly from the work that Anna and the whole traveler team had done on collecting and analyzing that data so looking at the different substitution rates and understanding. You know what which shopping trips would get convert converted to deliveries and which would be you know basically induced demand or extra trip.

When e-commerce increases all of those went into the specification of the household level e-commerce engagement model that we ended up estimating and incorporating into the Polaris workflow and there are similar findings around. You know I would say ride hail engagements that went into the beam model specification and others. So it's definitely the case that you know either directly or indirectly. Many of the findings from hole 12 went into the specification of the larger simulation studies and if you want to build on that especially for the bean case yeah and I will say I mean I think you summarize in terms of the work done in smart 1.0 that's an accurate description. I will say we're kind of trying to go even further in this next round of work that we're just now starting in particular in the B model there's a lot of emphasis being put into integrating some of the insights we gained from with regard to sort of the lifecycle factors and how that feeds through kind of vehicle choice and residence location choice and also mode choice with respect to kind of sensitivity of different preferences for people in different life cycle phases and patterns.

Great thank both of you! There's another question and I can't tell immediately which result is referring to. The question is the ride hailing scenario increased overall fuel consumption due to dead headings. Does it reduce overall cost given that customers do not need to own a vehicle? The phrasing implies that might be a modeling result but do either of you recall which result that might be referring to the right Austin one but it might be the workflow stuff I’m not sure that sounds like the right Austin. I think the workflow there was conflicting findings to that. So that some of the shared scenarios reduced overall energy consumption, so I would guess that's the right Austin finding yeah and I would say at least in that analysis I don't think they looked at total cost so I don't think I can speak to the exact answer to that question.

Yeah fair enough, I concur with that another question which is really to both of you and we'll start with Anna how do you study what people are going to do when faced with the technology that doesn't really exist yet so I guess Anna on your  survey work and empirical data work and Josh and the modeling and simulation side you're both looking at technologies that aren't you know prolific yet so how do you do that study?

Let's start with Anna well I mean that's the question isn't it! I think that you know anyone who's trying to study anything with regard to for example connected and automated vehicles is going to be facing the same question and the same difficulty the way that we've chosen to grapple with it in our case especially in the whole traveler study was to try to get at some of those underlying factors that that provide a deeper understanding of the underlying motivations and restrictions that could be translatable to context that haven't emerged yet and that's across the kind of the range of different technologies and services we're looking at, so from that it's not like we you know emerge with the final be all and undo single parameter value of the propensity to adopt connected motivated vehicles for example but that it you know contributes to a better understanding of what might matter and what might not and then that coupled with you know increased results from other survey work and field experiments and other things over time. The idea is that these models get tuned up and tuned up and improved upon more and more over time and you know we just all do the best we can. I think yeah I would I would generally agree with all that and say you know a lot of the work and a lot of the focus is really just on learning from analogs and under extending what we already know about how these behaviors go into these assumed future technologies and there are a lot of analogs out there and a lot of different data out there when we're talking about something like an AV for example it's a very new technology but it's not fundamentally something different than you know it's basically an uber with no driver. Right so it's there are definitely a lot of studies and a lot of data being collected around this topic that do allow us to at least make some educated guesses about how people will behave in these in these circumstances.

Thanks both of you for that I have a question on the whole traveler survey so I think this is for Anna the results presented on e-commerce from whole traveler were interesting but it seems like there would be a lot of variation across different households. Did you look at this variation? Yes in particular what we looked at is variation across households with respect to whether they had young children at home and also whether with respect to household income and you know the that we do reference in the slides that Josh showed on the references. There's a journal article where you can go into more detail on those sets of results but one of the interesting takeaways we kind of got from that level of the analysis was basically that you know when we look at those with higher incomes or those with children which we might think of as those with higher opportunity costs for their time or higher kind of demand for their time we don't necessarily see that e-commerce is replacing more shopping trips but what we do see is the likelihood that e-commerce is replacing more other types of time-consuming behaviors like cooking at home for example.

Okay thanks I might throw my own question here to Josh to kind of follow up that on that how do you in the in the agent-based models, how do you account for that heterogeneity across different households in the model? Is that is that taken into consideration? Yes, when the model was estimated we did look at all these different factors you know some of the obvious ones like income level. But some less obvious ones like their accessibility to take transit, their accessibility to nearby retail. So really looking at more of the geographic distribution and the transportation system distribution and incorporating that into some of the model estimation to really and that's tying back in this e-commerce behavior into the larger you know transportation system effects. We're trying to model and especially that we're trying to change through different scenarios. So we did capture probably not as much of the heterogeneity on the individual level as what was being explored on the on the whole traveler side, but we did capture some of that in terms of you know the household size the income level and so on.

All right thanks! Maybe time for one more question. Here, there's a question that asks: “have you done similar studies on for people living in rural areas?” So I’ll add to that, you know and it was whole traveler behavioral study results came from the San Francisco region a lot of the Polaris results and the beam results came from the Chicago region and San Francisco respectively. So what about studies for people living in rural areas? Maybe start with Josh on this one. So rural for sure is definitely included in the studies that we run. So when we say Chicago area it's Chicago but it's also about I think ten thousand square miles around Chicago’s metro model goes up you know past Kenosha in Wisconsin past Rockford and Illinois and past I think south bend or Michigan city and Indiana.

So it's a very wide geography that encompasses all different urban forms and some of the results we presented today were coming out of our Bloomington Illinois model which is sort of a rural to small townish area in Illinois. So we do try to apply these across different region types just to capture geographic heterogeneity and you know the way people access take shopping trips socialize. For example do tend to be quite different in different geographic contexts. We want to make sure that's representable in in the workflow so that's why these models tend to cover a wide region and then there is ongoing efforts as a part of the smart mobility program to extend these models to different areas of the country. So we're not just focusing on two cities we really want to capture a broad swath of the country when we're doing analysis like this.

All right I’m showing four o'clock eastern time, so I’m afraid we're going to have to end there. Again thank you to Anna Spurlock from Lawrence Berkeley National Lab and Josh Auld from Argonne National Lab for being our technical speakers today and thank you to everyone who joined us and who submitted questions.

I’ll remind you if you want some more information about the results and findings from Smart Mobility you can download our six capstone reports at the link shown on your screen here. You can also download presentations and videos from previous webinars, additionally at that same link you can register to attend our final two webinars one is two weeks from today focused on moving goods in the smart mobility system and then two weeks after that we will wrap up the series looking at EV charging infrastructure in the smart mobility system. So again thank you for attending please read the reports and please register for our future webinars thank you everyone and stay safe!