Hello and good afternoon. I will remind you all that this Webex call is being recorded and 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 use a video connection, you are presumed to consent to recording and use of your voice and image. I will also remind you all that we have muted all participants from our end and so there should be no issues there. And with that, we'll go ahead and get started.
Again, good afternoon. I'm pleased to welcome you all to our second in a series of webinars to highlight the work, the results, the findings, and the insights from DOE's Smart Mobility Lab Consortium. My name is David Anderson. I am the Program Manager for Energy Efficient Mobility Systems or EEMS. EEMS is 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 major research effort of the EEMS program and is focused on. One: understanding the system level impacts that emerging transportation technologies and services will have on mobility. And secondly: identifying solutions that improve Mobility Energy Productivity. Which is the topic of today's webinar.
These webinars take place every two weeks. they started on September 24th and will continue through early December. Each one focuses on a different aspect of what we accomplished and what we learned. The first phase of our Smart Mobility effort over the last three to four years. Today, as I said, our focus is going to be on the Mobility Energy Productivity or MEP metric - a new mobility metric developed through the Smart Consortium that considers the energy, the time, and the cost associated with the transportation system. But before we get into the details of the Mobility Energy Productivity metric, I want to give you a little bit of background; a little context on the Consortium. Some of which may be repetitive with my comments from the first webinar, but we do have some new people joining us today so please bear with me.
So EEMS is one of several research programs within DOE's Vehicle Technologies Office or VTO. Now, VTO has a demonstrated record of success in researching and developing advanced vehicle technologies that improve the fuel efficiency of our light duty, our medium duty, and our heavy duty vehicles. Traditionally, this R&D was focused on vehicle power trains and component technologies. EEMS and Smart Mobility really extends VTO's reach to the systems level to the right side of this slide. We look not only at connectivity and automation - at the vehicle and component level -, but we also look at how vehicles interact. How they 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 work to understand their implication in the context of an entire urban or metropolitan area. And that includes passengers and freight and not just on-road vehicles, but the entire mobility system which includes transit and micro mobility and pedestrians and biking.
DOE's mission is to accelerate the science and technology that drive U.S. prosperity and security, and along with our National Labs, we have a critical role in realizing the potential of Smart Mobility. Across our lab complex we have unparalleled computational tools, we have scientific and engineering expertise, and we have R&D resources that deliver real impact by creating solutions that address our nation's most complex and long-term transportation challenges. The opportunity space for transportation is much larger than any one organization, so we decided to take a multi-disciplinary consortium approach to mobility. We convened five of our leading National Labs doing transportation research - and this includes dozens of our most talented transportation scientists - and engineers and we built the Smart Mobility Lab Consortium - a multi-year multi-laboratory collaborative dedicated to further understanding the energy implications and opportunities of advanced mobility solutions.
So we can look back over centuries at how civilizations developed and how our cities were built. They were built along rivers and railways and roads. In other words, they were built near the ways that people and their stuff moved. This tells us that mobility is the foundation of how humanity interacts and connects. Today mobility gives us access to opportunities like healthy food, quality education, good jobs, adequate health care, and plenty of leisure and recreational activities. And so, by understanding the transportation system as a whole by developing solutions that address the system, not just individual parts, we have the potential to not just move Mode But to move smarter. And smarter means more affordably, more efficiently, and with more choice. And that means providing more access to more opportunity for more people.
So speaking of accessing opportunity, we use the term mobility quite a bit and it may mean different things to different people. So I want to give you our definition. So in the Consortium we define mobility to mean the quality of the network or system to connect people to goods, services, and employment; the things that define a high quality of life. We believe this is a comprehensive definition that captures the aspects of the transportation system that we're trying to understand and improve. So now we need to quantify it. Otherwise, if we want to improve mobility we need to be able to measure if we're successful. So we looked across the transportation landscape and found there were no existing mobility metrics that adequately serve our needs. Current transportation metrics tend to focus on utilization of the road network, for example. Accessibility metrics are usually single mode. But we're not just interested in improving the road network or focusing on just a single node, we're taking a multi-modal approach to connect people to the things that they need better. While at the same time, minimizing energy consumption. So we define this metric - we call it Mobility Energy Productivity or MEP - to quantify mobility in terms of the energy, the cost, and the time required to access opportunities and in just a minute Venu Garikapati from the National Renewable Energy Lab will describe the logic, the thinking, and the application behind this metric.
One of the things that Venu will show us is how MEP is calculated. So I realize none of you wants me to give you a math lesson and Venu will do a much better job of explaining the calculation behind MEP than I will. My summary of the equation is this, MEP is the cumulative utility weighted opportunity space for a geographically defined area. So utility weighted. What's that mean? That means there's a utility function that defines some sort of benefit. So this function is shown on the left here and it has three terms: E, T, and C. That's the energy, the time, and the cost for accessing all the opportunities available by all transportation modes. Again, Venu will provide much more information on this. I think it's useful to see how the equation works graphically or geospatially and so here we'll use Denver as an example. On the left you see a heat map representing the values for all the areas around Denver, Colorado for the car mode only. So there's a lot of green here and not so much red. Green means a higher MEP number. You can access a lot of opportunities around Denver pretty easily by automobile. In the middle, you see the MEP heat map for transit, biking, and walking modes combined. Now the scale is slightly different here, but not by much. Accessibility by these modes is really concentrated in the dense urban area. Further out in the suburban and rural areas, you have much less access in a given amount of time if you're using these modes. Now on the right, we have the heat map for MEP for all modes combined. Again the scale's a little bit different here. It goes up to 500. So it's hard to compare directly with the other maps, but it's it's pretty clear that mobility improves and MEP goes up when you have more affordable, more efficient, and more convenient choices.
So now I'm going to turn it over to Venu from the National Renewable Energy Lab to describe the logic and the thinking and the application of the Mobility Energy Productivity metric and to potentially correct any confusion that I've caused in my introduction. And so when he's done we'll have some question and answer period. And I'll remind you please type your questions into the Q&A box that you have on your screen. So Venu, feel free to take it away.
Thanks for ssetting up the context, David. I will begin with why we developed this metric. I mean David kind of stole my thunder a bit here. A couple of years ago - maybe more than a couple of years ago, when we were looking at transportation efficiency metrics that can be utilized in a variety of scenarios to measure the improvements or the lack thereof when technologies change, build environments change, or social demographics change... What kind of metrics there are to capture such changes? We ended up with two streams of metrics that exist: the first being transportation performance metrics that measures the utilization or efficiency of the road network. These are measures like vehicle miles traveled or volume to capacity ratio. These measures do a fine job at depicting the aggregate level system performance, but when we want to delve into the nuances of which specific technology has caused a shift in VMTR volume to capacity ratio it gets kind of difficult. On the other hand, we landed on accessibility metrics which measure the access to opportunities, but we found that existing accessibility metrics to a majority degree either are unimodal or unidimensional meaning they either measure access to one access using one type of mode or access to one type of opportunity. For example, access jobs or access to health care; things like that. So we quickly realized that a combination of these facets is required to answer questions that deal with various technology changes, infrastructure changes, or "How does emerging mobility technologies influence a community's overall mobility?" So we set our objective as how do we quantify the efficiency of a network or system to connect people to goods, services, and employment that define a high quality of life. That is the beginning beginnings of the Mobility Energy Productivity metric.
We've scaned the literature to see if there are relevant metrics that can educate us in the development of the MEP metric. And there are many metrics that some of you might already know. Metrics such as, walk score, bike score, or transit score look at access from a given location using the specific mode to a number of opportunities. If you take something like walk score, a walk score of 100 defines a location as a walk haven. Meaning you can reach many things using walking in that location, but walk only. Similar things exist for bike and transit, but there is not a lot of research on combining the modes or combining the different types of activities you can reach using each of these modes. So the job we had cut out for us is, "How do we effectively combine different modes that provide accessibility in a given location into a holistic metric and while doing so, in addition to travel time - which is used as the efficiency measure in a majority of accessibility literature - how do we incorporate energy and cost parameters into the computation of the metric?," because when we bring in the model component into play it's not just the time efficiency that plays an important factor. Energy efficiency and cost efficiency are things you'd like to consider too in depicting how future modes might impact mobility and energy. So the pseudo equation we've set up for the Mobility Energy Productivity before putting the math behind, it is that the MEP metric should be a function of mobility weighted by energy, cost, and trip purpose. Meaning what types of activities you're accessing.
Like any good research endeavor we've defined a few properties for the metric that we'd like to achieve before jumping on developing the methodology itself. And the properties we define for ourself are that the metric that we come up with should reflect the efficiency of accessing a variety of good services and opportunities. That it should be based on established accepted research - we've built our metric on the foundations of accessibility theory. So we didn't find a necessity to reinvent the wheel. We just took an existing wheel and tried to make a better wheel. That the metric that we come up with should be - should have the capability to be applied to any mode, existing, emerging, or future modes. That it should be determined by the three dimensions to begin with energy, cost, and travel time as David has already mentioned. And that it should be spatially scalable. Meaning that it's not just the score for a location but how do you aggregate that score to a city level that can be used as a barometer to see how the MEP is increasing or not increasing as time progresses or as built environmental technology changes. We developed the metric to be data agnostic. This is a very important facet of the metric that has helped us switch in or switch out a variety of data sources in the past couple of years to compute the metric. And finally the point of developing this metric is that it should provide a lens to compare the Mobility Energy Productivity of score of two locations within a city, how do any how do the MEP scores change between two planning strategies, or two technologies things like that.
The data spectrum that drives the metric comprises of five core elements. I'm not going to go into the details but I'll just go from the bottom to the top. We start with Travel Time. From a given location, Travel Time is the first thing you use to quantify how far can you reach and then go to the Land-Use layer. Meaning once you know how far you can reach we quantify what you can reach within that Travel Time right. The Cost Measures the Energy Efficiency Measures and the Travel Demand Data are all parameters that either weight or de-weight the opportunities that can be reached from a given location using a given mode.
For folks who are not as aligned with the concept of accessibility theory or travel time based research, I'll first define what an isochrone is and then go into the methodology for developing the metric. An isochrone is defined as a line drawn on a map connecting points at which something occurs or arrives at the same time. Simply put, an isochrone is if you started the location of the blue pin, in a given amount of travel time, how far can you reach in 360 degree dimension is what an isochrone depicts. So we draw for any given location an isotone for a given amount of travel time and then quantify all the types of opportunities you can reach from that location within that travel time using that mode. What type of opportunities you might have? We define different types of activity categories like educational opportunities, grocery stores, hospitals, etc. We do this not just for one travel time, but for each mode we do this once in 10 minutes and then increase our isochrone boundary to 20 minutes. So when you increase your travel time from 10 to 20 minutes you can travel a little further - 30 and 40 minutes. So within each of these ten minutes, we quantify how many incremental opportunities can you reach if you increase your travel time by that much. Once we quantify the number of opportunities that people can reach within a given amount of travel time using the given mode - that mode can be any more like a car, transit, biking, walking, etc. - this opportunities measure is weighted in three dimensions time, energy, and cost. But there is another important facet that we brought in to reflect the frequency of activity engagement. This is a parameter that we brought in a little later in the development of the metric, but one that has helped us reflect the metric in the context of how people engage in different activities that exist.
Coming to an illustrative example on how we compute the MEP metric. For any given city, the first thing we do is we slice the geography for which we are computing the MEP scores along the latitude and longitude into square kilometer pixels. So for each pixel then, we start querying in three dimensions: The first dimension is Time. In 10, 20, 30, or 40 minutes, how far can you get from this location? The second dimension is Mode in 10 minutes by car or in 10 minutes by bike or in 10 minutes by transit. How far can you get from this location? And the final dimension is Activity. In 10 minutes by a car, how many jobs can you reach; or in 10 minutes by bike, how many restaurants can you reach is what we quantify. Once we do this repeated querying along the Time, Mode And Activity dimensions. We go into a dimension reduction process which starts with proportioning the activities by what we call an activity engagement frequency. We brought this factor in a bit later in the process to reflect the fact that it's not just the availability of an opportunity within a given amount of travel time, but we all engage in different types of activities to different degrees. Employment is a more regular activity. We go to access jobs more frequently than other types of opportunities such as going to the restaurants or going to the movies. So once we quantify how many number of things you can reach within a given amount of time and using the given mode, we proportion the activities by what I call the rhythm of life factor. Basically, at what frequency do you participate in the activities that are reachable within a given amount of travel time from a location.
Next we go into the dimension [inaudible] for Time. The reason behind doing this is the farther an opportunity is from you the less attractive it becomes to you. This is the foundation of accessibility theory travel [inaudible] says that the same opportunity if it's x minutes away it's y percent less attractive in for ... In a similar way we do energy and cost weighting. Which means that if a mode is more energy intensive in getting you access to the same number of opportunities, that mode gets de-weighted than another mode that's more energy efficient. As an anecdotal example, if you can reach 100 things using car in 10 minutes and if you can also reach, hypothetically, 100 things using bike, then the bike mode will get weighted higher because that's a more energy efficient mode compared to car which is not as energy efficient as a bike. So this is the equation that David has already shown. I probably won't go into the details of all of the parameters in the equation, although, I'll mention three key things in this equation. The first thing is Oijkt, which is the opportunities measure. This is the raw number of opportunities that we count. Which I've shown in the slide before. The second thing is frequency of activity engagement. So once we determine the number of opportunities within a given amount of travel time, we do this activity engagement frequency proportion. The final thing is a weighting factor which is eMikt. This is the utility weighting that David has mentioned which includes energy, time, and cost parameters. For now, we are looking into adding additional parameters that pertain to different modes into the utility weighting as extensions for the [inaudible]
Putting some numbers behind the parameters that I've shown. These are the numbers that we use for energy intensity and cost for various modes from citable sources like the transportation energy data book or other published literature. We use these numbers to apply the metric to compute baseline MEP scores across a number of cities.
That brings me to the next part of the presentation. We've talked about how we've developed the metrics, now I'm going to talk about how we implemented the metrics. And I'm going to talk about two parallel implementations of the metric. The first is a standalone application of the metric, which is taking the methodology we have using publish using publicly available data sources - or some data that's available to us through data partnerships with some entities - and applying the metric to compute baseline scores today for any given city. Columbus, Ohio is one of the first cities where we computed the MEP score. As David has alluded to before, green in the map means that from that location you're able to reach a greater variety of opportunities using more number of modes in a more time energy and cost efficient manner. A red colored pixel, on the other hand means, that you either cannot reach as many opportunities from that pixel or you cannot reach those opportunities in a time, energy, and cost efficient manner. You see this gradual transition from the downtown to the suburban location from green to orange to red? This is just a facet of the density of development in downtowns meaning you can reach a greater number of opportunities within a short amount of travel time if you're in the downtown versus a suburban location. One thing that kind of caught our eye when we first looked at the map for Columbus, Ohio is this red patch in a gradual transition. We went back and looked at why a red patch exists and found out that this is actually a rock quarry where you cannot travel using a car, bike, or even walk. So that led to a very low MEP score for that location.
Depicting geospatial scores is a good way to identify the differences in MEP scores across various locations within a city. But when we want to track the metric across time for the same city or if you want to apply the metric in different scenarios, just changing our explanations or observations on the visualization might be a bit difficult. So we thought we'd come up with a waiting schema where the metric can be aggregated to any geographic level of choice we've chosen the city boundary as the aggregation level here. And this slide just shows a couple of examples where if the pixel level MEP scores are exactly the same, but if the zonal populations - which totally about 900 people - are spread differently in two different examples here that will lead to widely different MEP scores. Why I bring this up is it's not just how high a Mobility Energy Productivity score or location has, but how many people are able to exercise that Mobility Energy Productivity is is of great importance. So we do population waiting to depict that if a greater proportion of the city's population is able to exercise a greater Mobility Energy Productivity in that city then the city level MEP score comes out to be higher.
The population density weighted metric for Columbus, Ohio turns out to be 198. So as I mentioned, we started working on Columbus a couple of years ago. We've greatly increased our efforts in computing the metric for a variety of cities since then.
The metric I've shown you the dimensional map across time, energy, and cost domains. We compute the metric elementally from each layer. So each time, mode, and activity layer. So it's easy to peel the layers back and look at MEP scores for any specific dimension of interest. Here I'm showing the mode dimension where driving, transit, bike, and walk are shown separately. David has shown a similar map for Denver with transit, walk, and bike combined. You can see from here that the driving map is green, not because driving is the most energy efficient mode, but you start from anywhere in Columbus within 40 minutes you can cover a large geography within Columbus. So the time dimension is actually favoring the drive Mode As opposed to something like transit which might be limited by schedules or limited by routes. So transit doesn't get as high a MEP score as other modes here in Columbus.
Over the past couple of years, we were able to exercise the metric for a number of cities within the United States to bring geographical equity. We thought we'd compute the metric for the most populous city in each state and then add a few other cities of interest. If we take large states like California, Texas, or Florida, there might be more than one city that might be interesting to look at. So we first computed the metric for the most populous city in each state and then went ahead and computed a few other cities of interest. In total we have computed the metric to date in about 108 cities.
Usually when I present about the metric in any conference or venue I give the local example. So if I present this in DC, I'd take an example of what the MEP score is for Downtown DC or something like that. Because this is a virtual conference, I thought I'd give an example of the location from where I'm presenting. I'm presenting here from Denver, Colorado. And the example I'm going to give you is our MEP score three weeks ago 173. We used to be in a location that has reasonable access to a variety of restaurants, grocery, stores, hospitals, etc. We moved just in these three weeks to a location that has lower MEP score. Now you think with all of the research that I do and all of the great work that our team does, I should be able to convince our family - my wife in particular - to move to a location that has a higher energy score. I'm somehow not very successful in doing that. She valued being closer to the mountains, for folks who know the region, over having a higher energy score. So here we are.
Computing the metric in a baseline senses only have the fun. The real advantage of having a metric like this comes in applying the metric for a variety of scenarios. So early on in the development of the metric, we've exercised the metric to depict an illustrative scenario where we said, "In Columbus, Ohio is in a future state the travel efficiency of the network state remains exactly the same, but if energy efficiency of travel drastically improves by 200 percent, how would that get reflected in the MEP calculation?" And you see that from the left to the right I'm showing this to the same scale here so you can identify the difference. The map cleans up, not because there is great travel time efficiency, but because there is great energy efficiency. We should be able to reflect changes in travel time efficiency just the same or if travel and energy efficiencies remain just as there today, but cost of travel reduces, we will be able to depict that as well. The caveat, however, for this scenario analysis is that it does not account for any secondary effects of MPG increase. If there is more energy efficient vehicles available, people might choose to travel more. That might lead to network condition and some adverse impacts that might in turn lead to reductions in Mobility Energy Productivity scores. The metric calculation itself will not be able to capture such secondary effects, but such effects can be captured by linking the metric to advanced simulation tools that capture such nuances.
We did just that so the second stream of application I am going to talk about today is the MEP application in integrating the MEP calculation with the Smart Workflow Modeling Process. Some of you might think... sorry for my seasonalities here... some of you might have attended the previous webinar where [inaudible] regarding the Smart Workflow Modeling Process which is a comprehensive modeling suite comprising of energy simulation models, transportation simulation models, land use models, etc. all working to run scenarios that are set 15 - 20 years into the future to see how the mobility and energy outcomes of those scenarios will be. The MEP calculation sits at the end of the Smart Workflow Modeling Process and captures the outputs of transportation simulation models, energy simulation models, language models, etc. and digests all of that information into a score that can be used to see if a particular technology or a particular language change has led to MEP increases or decreases when compared to the baseline scenarios.
A difference I'll identify here which relates to the data agnostic nature of the metric I've mentioned before, is all of the standalone application I've presented before uses third-party data sources or publicly available data sources. The calculation procedure is exactly the same, but when we integrate the metric with the workflow, we flip the data sources from third-party publicly available data sources to data sources that are within the Smart Workflow Modeling Process. By doing so, we are able to develop just one model or one metric that works with any type of data. That's an advantage that has helped us connect with two distinct workflow processes. One developed out of the Lawrence Berkeley National Lab and one developed out of the Argonne National Lab. Both of which are going to get presented next week by Zach Needell and Josh Auld.
To give you a sense of how the metric works when we connect this with the workflow modeling process, I'm showing some sample outputs here from San Francisco. This comes out of the workflow modeling process that's run out of the Lawrence Berkeley National Lab and the beam transportation simulation model is at the center of this workflow. You can see for San Francisco the score here is computed using the outputs of all of the simulation models within the beam workflow. Similarly for the Chicago side of things Argonnne National Lab with Polaris tool has... Polaris Transportation Simulation tool at the center of the workflow. They've run a variety of future scenarios and we ingest outputs of all of those scenarios to compute the MEP scores. So this shows a sample output for Chicago. I gave an illustrative example on how we applied the metric to capture energy efficiency improvements in Columbus, Ohio with the caveat that in that instance you're not able to capture any secondary impacts of energy efficiency improvements. Here is an example that overcomes that limitation when we connect the model to workflow modeling process. I'm using an example that compares two modes here; Mode A and Mode B. Both are road modes. Meaning they both travel on the same network so they both have the same network speeds. But let's say Mode A, you don't have to wait to use that mode. Mode B, you'll have to wait for about five minutes to use that mode. So here Mode B is slightly time inefficient compared to Mode A. If you look at both the maps you'll be... maybe it's difficult to do this at first glance. To me it's a bit easier, because I've seen a few thousand MEP maps over the past couple of years. But Mode B gets a slightly lower Mobility Energy Productivity score and to show that, we've aggregated the score to the city level. You see that the overall MEP at the city level for Mode B is slightly lower than Mode A primarily because of the time inefficiency here. If we compound the time weighting with energy weighting, and here I am implementing energy intensity weighting. Let's say Mode A and Mode B have similar energy intensities - Mode B might even have slightly better energy intensity on a per passenger mile basis compared to Mode A - when you compound the time and energy rating, you see that Mode A still comes out on top. So even with slightly higher energy intense... slightly better energy intensity for Mode B, it's not able to overcome the time efficiency that Mode A has when we did this with just the time weighting factor.
Finally, if we bring in the cost weighting and say that for traveling in Mode A you'll have to expend about 20 cents per passenger mile and for Mode B if you have to spend an [inaudible] higher, now you can actually visually discern the difference between Mode A and Mode B. With Mode B snap being much more red compared to Mode A. When we aggregate this to the city level, you see that the Mobility Energy Productivity of Mode B is about half or even less than half compared to Mode A. So this takes into account how people are using these modes, what are the time energy and cost impacts of using these modes to travel to various locations within the city. Integrating this with the workflow modeling process gives us the flexibility to implement the changes in any one of these dimensions or a combination of these dimensions.
So the example I gave was an anecdotal one. What if we implemented this for a variety of scenarios that looked at short and long term futures? The workflow modeling process did just that where a scenario was done for a short term future with high sharing and partial automation. And then two long term scenarios were run. One for high sharing and high automation. In this scenario, folks are allowed to own their individual vehicles as well, but to a decreased degree and there is a high prevalence of sharing. Alternately, if there is high automation but folks are not willing to share, there is another scenario run parallelly with the high sharing scenario... with the high automation scenario but with low sharing. Both Josh and Zach are going to delve much deeper into the details of the modeling processes for these scenarios next week for both of the workflows that came out of their labs.
I'm going to look at some of the results from an MEP lens. So for some similarities and differences, you'd expect differences to exist. Both of these models are implemented in different geographies. And both these models consist of components that make different assumptions. But the similarity in business we saw was, no matter what modeling workflow we saw that the Shared-AV scenario led to the highest MEP score compared to the baseline scenario in both the beam and the polaris workflows. This goes to show that the metric can be used, data agnostically and across the workflows, to tell to provide a common lens to identify the changes that will happen in technologies or in time energy and cross domains across a variety of scenarios. You'll see that the order of magnitude difference from baseline in beam is slightly it's definitely lower than that of polaris. The reasons for that in discussions with each of the modeling teams we found is that the MEP improvement in the private and the Shared-AV scenarios definitely exists in the San Francisco beam based workflow, but to a lesser magnitude; because increased congestion in the San Francisco implementation of the workflow offsets some assumed vehicle efficiency improvements in the long-term future. If the improvement is difficult to discern visually here I've shown the cumulative distribution function. If the cumulative distribution function moves to the right, that means that there is an improvement in the MEP score and in this case between the Shared-AV and the Private-AV scenario. We see that the Shared-AV scenario has about a 13 higher MEP score than the private event scenario.
Similarly for the Chicago workflow, we do see an increased order of magnitude difference for the Shared-AV scenario. This, I think, totals about 67 percent improvement from the Private-AV scenario; and this is because in the Chicago implementation of the workflow - which again Josh will go into much more detail on - we saw greater magnitude improvements in the MEP score going to a combination of decreased congestion coupled with assumed vehicle efficiency improvements in the long term.
Coming to some next steps on where we want to take this metric, we've just finished the first iteration of the Smart Mobility Consortium and have kicked off the second iteration at the beginning of the fiscal last week. So we are continuing the MEP research into the second iteration of the Smart Mobility research. A few enhancements that we thought we'd address right off the bat are including multi-modal isochrones right now we develop an isochrone for a mode separately. So we either have a walk isochrone or a transit isochrone or a car isochrone. We want transfers between modes and develop multi-modal isochrons where you're traveling to a transfer stop in your car - taking the bus or the train - and completing the final part of your journey using [inaudible]. I've shown the MEP score as a single metric for a location. We want to move from that to depicting the MEP score as a range. This will give us the flexibility to identify if the MEP score for a location is higher in certain times of the day versus others. We also want to bring in individual social demographic characteristics into the MEP calculation. So for the same location the MEP score might look different to a car friendly person versus a car adverse person. The MEP maps I've shown today are kind of one-off maps that we as researchers develop to show in presentations and whatnot. Through the course of the smart 1.0 modeling process, we've identified that it would be beneficial to put these results in an interactive dashboard that can be made available to stakeholders so they can look at various future scenarios and see how the MEP improvements happen with changes in time, energy, or cost factors. Finally, some work that's already underway is I've shown some scenario... examples and analyses for San Francisco and Chicago which were predominantly done in the Smart One workflow modeling process. Over the past quarter or so we've... both the workflows have increased their footprint in applying the workflow across additional cities. So we now have, through the workflow connections, computed the metric in Austin, Detroit, Atlanta, and more cities to come in the future.
Last, but definitely the most important, this is not the work of one person. I will end with my wonderful team who make all of this great work possible. Ambarish, Chris, Rob, Stan, Yi, and many others who have contributed to the work I presented today. With that I'll pass back to David.
Thank you so much Venue. Yeah, great work both by you and by the team.
And now we will move into our question and answer period. Again I will ask you all to submit your questions via the Q&A box that's there on your screen. We will triage them as they come in. We already have a number of questions and I will start going through some of them.
So one thing that's [inaudible] question is very pertinent and timely, I will point out that much of this research has been done again over the past three to four years prior to our current transportation situation which is, you know, obviously COVID induced - quarantine requirements and things like that. So we are ironically not as mobile as we were before. And so the question is, you know, given where we are today in terms of a lot of telework and, you know, virtual meetings rather than physical mobility, how will the MEP... how will the Mobility Energy Productivity metric be adapted to the to the current situation? How can it be used to look at outcomes - whether short-term or long-term - in terms of our transportation limitations due to COVID.
Thanks for that question David. I'll answer that question in two parts.
The first is if I was asked this question six months into the research on the MEP metric, I would have said there is no way we can capture the changes on when people won't travel anymore. But through the course of our own peak into the literature or discussing amongst the team, we've identified that it's important not just to quantify what physical things exist within a given amount of travel, but it's extremely important to identify how people are engaging in activities that exist around them. That's the frequency of activity engagement factor. So in a situation as you mentioned now, I've given the example in the presentation of... well jobs are more regular activity. We are all still fortunate to engage in our employment, but we are not doing so in place of employment. I'm presenting from my home location here and I'm sure the the case is similar for many of our audience here. So when we want to apply the metric in the current situation, we will not change our calculation in terms of what we quantify - that part will still remain the same - the activity engagement frequency is what instead will get introduced into the metric to say, "The Mobility Energy Productivity has changed for a specific type of activity because this activity is not being accessed as frequently."
The second part of my answer is, we are in a situation where we are not accessing activities as much. It has certainly improved from the state home orders back in march or april. But we are still not accessing opportunities as much as we used to, but we are still receiving goods delivered to us, right? So when people are not accessing things, but when things are being delivered to people that's an other facet we thought about in the development of the MEP metric. And we've parallelly initiated the development of a freight version of the Mobility Energy Productivity metric. It does the counterpart calculation of the MEP equation here where it quantifies the efficiency of how things can reach people in a given location as opposed to how people can reach things. So that work is very much in progress. We've published some work on it very recently, but we intend to extend that into smart... to our other activities responding is available.
Great! Thanks for that Venu.
A couple questions here. One of which came through the the chat box, but there's a similar question in the Q&A box. So I'll remind you to enter your questions in the Q&A box. About, you know... Are there any extensions to MEP to address different mix of powertrain types? So I guess the question is, how do you look at, for example, combustion powered engines versus EVs, etc., fuel cells?
Sure! Yeah. That's a great question.
The energy intensity part of the MEP calculation right now is at the more level. So for a given mode - let's take a car. We look at the fleet mix of the city and come up with one energy intensity factor, for a car, for the whole city. Now if the fleet itself changes over time, that gets reflected in overall energy intensity improvements for that mode within that city. It's easy to disentangle that calculation and say, "For the car MEP calculation, we can do an internal combustion engine based MEP calculation... MEP based car calculation.", but I think that will lead to similar or the same result in terms of like combining the energy intensity and applying it as one factor versus disentangling the MEP calculation and aggregating it to the mode. So we have the architecture to do it. The way it happens now is we aggregate all of the energy picture of a mode in the city to one number.
Right and I'll add to that - and maybe this is repetitive - you know, though in the workflow when we look at those future scenarios that you described, there are many many assumptions that go into each of those future scenarios. One of those assumptions is related to fleet mix, what the powertrain mix is for the different on-road vehicles - whether they are passenger vehicles or crate vehicles. And so... and so yeah one of the outputs that comes from our modeling that is an input to the MEP calculation is essentially the energy intensity of the modes and that's based on, you know, the modeling that we've done the assumptions of the fleet mix and the power train technology. So but as you said it's aggregated to an overall energy intensity number apparently.
Okay. Hopefully that answered the question.
Another question is, are there case studies or examples of how a particular city or a planning organization has used the MEP or is interested in using the MEP to evaluate transportation and or mobility investments or make some of their planning decisions? So, yeah, I always think this is a great question ultimately our tools that we create we want to deploy. We want to put in the hands of the stakeholders that will use them. So how might transportation authorities or, you know, decision makers use the MEP metric to plan investments or make their decisions?
So the first part of the question David is, what we are planning to accomplish by the end of the Smart Tool research activities by engaging with many cities and presenting a variety of future scenarios that can lead to MEP improvements and then suggest which types of technology improvements or build environment changes can lead to those improvements. I will probably let you speak to that a bit more. But in terms of how cities are interested, case studies with specific implementations do not exist yet, but interest from cities is actually there and quite a bit. We are engaging actively with the Colorado Department of Transportation the Delaware Department of Transportation and, through the Florida International University, the Florida Department of Transportation to integrate the MEP calculation the same way we integrated it with the workflow here, but with the travel and transportation demand models for each of those states. So we are in various stages of discussion with each of these entities to be able to provide MEP as additional efficiency measure to evaluate various infrastructure and technology investments.
Okay Thank you. Yeah, I will not add to what you explained there, because we have a lot more questions coming in.
One question... the question... is did you consider with your models the impact of MEP score on street routes with one-way routing in the same city with two-way routing for starting and arriving at the same locations?
The I will defer to you for the answer, but I think that is an exercise that is beyond MEP that would involve our modeling workflow to look at, you know, different scenarios in terms of, you know, if streets are one way or two way and then we can apply the MEP metric on top of that to understand the impacts, but I will let you take the more detailed answer... well give the more detailed answer.
Sure. In terms of the capability to do it, it certainly exists. We do not use any third-party enterprises to compute our travel time isochrones. So if a property for the link can be defined as if it's a one-way link or a two-way link similar to the illustrative scenario analysis I've presented here, we can just say that all the links in Columbus, Ohio can be made one way and what's the MEP outcome of that? That's pretty easy to implement now what happens when that... when such a policy is implemented, how will people travel or not travel on certain routes? That the MEP calculation alone will not be able to tackle. As David mentioned, that we have to run through the workflow modeling process and then feed that output into the MEP calculation procedure to depict the outcome of such a scenario.
Alright, thank you Venu.
This should be a quick question; fairly interesting question: How are the units of the energy intensity, the travel time, and the travel cost unified before they are used to compute the MEP metric? The MEP is unitless, right? So how do we go from these three different metrics or quantities of different values and combine them into one?
So the energy cost and time parameters in the utility equation are additively separable. Meaning there is an "alpha" times "e," a "beta" times "t" and a "sigma" times "c." If we did not use the weighting factors ahead of each of these parameters, then we'd have to worry about the units of energy cost or time. But by adding these weighting factors or the alpha, beta, and sigma parameters, they essentially come up... they essentially have the property that they have the inverse unit of whatever is the parameter that you're talking about. So when we look at cost as a cost per passenger mile, the sigma parameter that we multiply with the cost per passenger mile as the inverse unit so that would be passing the micro cost. So we when we do that multiplication the utility that we use for the weighting itself is dimensionless that gets applied to the opportunities measures. So what we end up with is a utility weighted opportunities measure that doesn't have any. And this is fairly common practice in various transportation and travel demand models where utility is measured as a function of a number of parameters that span social demographics or travel time measures [inaudible] environment characteristics. And the coefficients usually bear the inverse sign making it easy to combine various facets that define utility.
Right, great! Hopefully that answered that question.
Here's another question that - I know we're getting a little short on time, but I want to be sure we get to this one, because it's something that we've talked about quite a bit in internally. So the question is it appears that the value of all time is the same - so this is a value of travel time question - however travelers may enjoy spending an amount of time on travel with a positive value attached to it, but a negative when a trip or the total travel time in a day exceeds some amount or is too slow. So we've talked about this in terms of you know what's the value of travel time when you're driving versus riding in a fully automated vehicle and being productive in some other ways, but this sort of expands that question. So...
So all of the results that I've presented today we agreed as a consortium that we'd use the experienced travel time as the the quantifying measure. So a one minute of travel in a given mode actually means that you travel one minute in that mode. With value of travel time that one minute might not seem like a minute if you've traveled in an automated vehicle. So it's easy to flip that switch and move to a value of travel time based in MEP calculation. Actually we started there and moved towards the experienced travel time realm. So if all of the transportation simulation models that we are working with already provide value of travel time parameters in addition to the experience travel time values. So it's pretty straightforward in a future scenario in an AV if a minute of travel doesn't quite look like a minute of travel to be able to compound that effect into the MEP calculation.
Alright. I'm going to ask one more question as we wrap up here. The question is: What would be the time latency of the MEP calculations? So in other words, how often do we need to recalibrate the model in a certain location that...? I'll let you take that Venu.
So I'll probably give a more casual response than I should do this question. My response is that data is available. As I mentioned, some of the data is from publicly available data sources. So we are at the mercy of when that data source gets updated so to speak. For example we use loads data for employment and that might be available at different times of different types of time frequencies. Similarly, we use network data that's either publicly available or licensed. So if and when we have the data available is when we'll be able to re-up the calculation of this course. The 108 series that I've shown, as an example, we computed this course for those probably like, I'd say, three months ago. When we next computed is if there is a significant change in either the land use or the travel time impacts and if that data is available to us is my simple answer to that. For a city, I think if you're able to compute this course at a quarterly or a bi-annual basis that should be enough to track if here are any improvements or the lack thereof [inaudible]. Hope that answers the question.
Yeah, thank you Venu. There's a lot more interesting questions coming in, but we are at time and so I will invite you, if your question was not answered, feel free to submit it to the email address that you see there on your screen. EEMS@ee.doe.gov. Now if you go on to the next slide...
I want to remind you all that we have published a set of six capstone reports and we invite you all to visit the link that is shown there. Where you can download these reports and where you can also access registration information for future webinars, you can access the transcripts and the recordings from past webinars, and you can even download the presentations from past webinars. Again, the link is shown on this screen as well. Our next webinar webinar - number three - is scheduled for two weeks from today october 22nd. These first two webinars have been somewhat structural. So we talked about how we developed the modeling workflow and we talked about how we developed the Mobility Energy Productivity metric. The next several webinars will be very results-focused on what we learned from the workflow and the other research that we did and how MEP applies to it. So we invite you to visit the link register for future webinars. Again, thank you for your time and stay safe and have a great rest of your day. Thank you.