The U.S. Department of Energy’s Energy Efficient Mobility Systems (EEMS) Program envisions an affordable, efficient, safe, and accessible transportation future in which mobility is decoupled from energy consumption. The EEMS Program conducts early-stage research and development at the vehicle, traveler, and system levels, creating new knowledge, tools, insights, and technology solutions that increase mobility energy productivity for individuals and businesses. 

The Systems and Modeling for Accelerated Research in Transportation (SMART) Mobility Consortium (Consortium) is a multi-year, multi-laboratory collaborative dedicated to further understanding the energy implications and opportunities of advanced mobility solutions. The Consortium is the EEMS Program’s primary effort to create tools and generate knowledge about how future mobility systems may evolve and identify ways to reduce their energy intensity. It also identifies research and development gaps that the EEMS Program may address through its advanced research portfolio and generate insights that will be shared with mobility stakeholders.

The Consortium consists of five focused pillars of research: Connected and Automated Vehicles,

Mobility Decision Science, Multi-Modal Transport, Urban Science, and Advanced Fueling Infrastructure. This report was developed as part of the Urban Science Pillar that aims to evaluate the intersection of transportation networks and the built environment in terms of energy opportunities. 

Opportunity – niche

As municipalities have scaled back their smart city ambitions — from grand plans to more modest projects — a space has opened up for niche vendors. No longer the sole realm of big-name corporations, smart city technologies are increasingly being purveyed by niche suppliers.

One such vendor is AutoGrid Systems of Redwood Shores, Calif., whose products analyze energy data generated by smart meters, building management systems, voltage regulators, thermostats and other equipment.

The niche play fits into local governments’ overall plans. “The smart city evolution is happening alongside this evolution in the electric grid. So cities view our technology as a tool to integrate all of these sources of energy,” said Quique Schwarz, vice president of products and analytics at the 50-person firm.

As niche players, firms like AutoGrid promise to deliver a level of specific expertise a bigger provider might not possess. “We work with just this problem, so we are able to suggest best practices that we see globally,” Schwarz said.

AutoGrid is hardly alone in its efforts to carve out a space in the shadow of the giants. Consider the recent Smart City Expo World Congress in Barcelona, which drew some 450 exhibitors, many of them niche companies like BreezoMeter (air pollution); Atos Worldgrid (energy management); Dimenteq (GIS solutions); and Black and Veatch (engineering, procurement and construction services).

“There is this fascinating, complicated ecosystem developing, and it is the perfect time for these niche players to have a place at the table,” said IDC analyst Ruthbea Yesner Clarke. “The big vendors have the technology in general, but they may not have the specific expertise.”

Smart planning modeling

  • State of the Practice for Land Use Forecasting 
  • Theory and Data for Forecasting Land Use
  • Scenario Planning and Visioning: I-PLACE3S
  • Spatial Input-Output Frameworks: PECAS
  • Dynamic Microsimulation Frameworks: UrbanSim 
  • Modeling Real Estate Demand 
  • Modeling Real Estate Supply 
  • Scenario Planning and Visualization

IOT Hardware and software Vendor profile:
Tridium’s Niagara Analytics

Tridium’s Niagara Analytics Framework is a data analytics extension to the Niagara 4 Framework® included on Niagara 4 Supervisors and JACE® 8000 controllers (requires Niagara 4 version 4.2). It gives Niagara users the ability to apply a variety of analytic algorithms and diagnostics to both historic and real-time data available in Niagara. Users gain insight that helps them better manage their buildings and operations.

Type: Smart city analytics Product
Categories: Analytics, Asia Pacific, Building Automation, Central America, Data Centers, EMEA, Industrial, N. America, S. America, Stand-alone Applications

First introduced in the Americas in 2014 and available globally since 2015, Niagara Analytics is the only data analytics engine built on Tridium’s industry-leading Niagara Framework®.

Niagara 4.2 includes integration of Niagara Analytics 2.0, which takes the robust analytics capabilities of Niagara 4 to a whole new level. Together, these releases bring the benefits and efficiencies of data-driven performance to users.

Niagara Analytics 2.0 can be utilized locally using real-time data in an embedded controller for energy optimization, or for data analysis using historical information you’ve already saved, making your staff more effective and your buildings more efficient.

• Familiar Niagara Framework wiresheet programming
Real-time, on-premise analytic control

JACE® 8000s
• Supports real-time energy optimization

Automated control strategies
• Advanced alarming
• Fault detection and diagnostics (FDD)

Major considerations


Predictive Analytics have been used in several cities across the world to help predict where crimes are likely to take place through historical data and geographical data. These have seen significant success in cities like London, Los Angeles and Chicago. Through data, it is often not even necessary to make arrests, having police officers appearing in certain areas at specific times has seen crime rates drop.

Through data use like this, we could see a significantly safer cities, with police who can stop crime without needing to put themselves at risk of harm.

City Planning

When looking at the city planning function, analytics and data are not often considered. People still see their use in web traffic and marketing, rather than in the physical creation of buildings and spaces.

This does a great disservice to the power than data can have on anything from building zoning to amenity creation. It allows models to be built to maximise the accessibility of certain areas or services whilst minimising the risk of overloading important elements of the city’s infrastructure. In Short, it create efficiency.

Too often we see buildings being built in areas that seem suitable but can have a considerable effect on another area, without this being taken into consideration during the planning process. Using data and modelling it is possible to map the infrastructure outcomes of any use of space with a high degree of accuracy.


During the London 2012 Olympics the network needed to deal with 18 million journeys made by spectators throughout London. It was no fluke that the network coped.

The TFL and train operators utilised data and analytics to make sure that the vast majority of journeys ran smoothly. It allowed them to input data from events to predict the numbers who would be travelling and make sure that transport was running effectively to make sure that spectators and athletes could be effectively transported to and from the stadiums.

Through using data like this throughout a transport network, it will create effective and flexible public transport, decreasing delays and increasing efficiency. Using data to not only predict when peak times will be for upcoming events, but to help monitor equipment will mean that reliability will improve and accidents will decrease.

Future Proofing

Often when new areas are created or become popular, the infrastructure in place is not good enough to sustain continued growth, which can hinder further improvements in the area. Even basic amenities like water and electricity can be effected by a sudden influx of businesses or residents. Through the use of modelling and predictive analytics, it becomes possible for city planners to see where these areas of growth are likely to be and how large this increase will be. Amenities can then be upgraded to accommodate this.

In this way growth in certain areas can continue without the need for services to catch up.

Web Provision

The general gripe that many have with the idea of ‘Smart Cities’ is that governments or companies introduce fast internet speeds and then declare that because companies have the opportunity to access it, it is now officially smart. A smart city is not instantly made because people can get onto Facebook quicker or can instantly watch Cat videos.

Providing fast web access is one thing, but it needs to be in the correct areas and for the correct people. The ability to shift bandwidth within a city will be a key component to this. Knowing when and where bandwidth should be prioritised is a key part of this and data is the compass to help steer it in the right direction.

The basic premise is that bandwidth should be highest in commercial and financial areas from Monday to Friday and in more residential areas on Saturday and Sunday. But there is more complexity than this and the opportunity to maximise bandwidth down to much smaller scales, an area where data and analytics can play a key role.

For instance, if an area wants to attract more high-tech industries and web development companies allowing bandwidth to be higher in those areas is going to be important and data modelling will allow this to be done most effectively.


One of the keys to sustainability is monitoring and having effective controls in place to quickly make changes in order to keep output at a certain level. Data is the most decisive factor here, it allows for governments and companies to see how their outputs are having a positive or negative result on the city as a whole. Being able to check and control levels of pollutants can help with zoning, placing pollutants in areas of the city where they can do the least harm or helping them to reduce their harmful output.

This monitoring also creates the opportunity to see which technologies work best in reducing pollution and what new innovations could be used in particular areas in order to prevent further environmental damage.


Many cities across the United States seek to develop, evolve, and mature by having their Smart City approaches informed by data and models to manage challenges, opportunities, and uncertainties associated with shifts in mobility technologies, human behaviors, and policy strategies. This study aims to provide a foundational baseline cross-city assessment of current data and modeling environments of the U.S. Department of Transportation (DOT) Smart City Challenge finalists (referred to as “SCFs”) with respect to their ability to support critical analysis at the juncture of transportation and energy. Specifically, this effort focused on how urban and regional data environments and travel demand models (TDMs) are evolving to support and enable higher-level mobility and energy analyses, planning, and modeling, and the motivational factors driving cities to develop and enhance this analysis and modeling capacity. 

Particular emphasis was placed on planning for a future of automated, connected, efficient/electric, and shared (ACES) mobility transitions through the uptake of technology and infrastructure transformations. While travel demand modeling has seen consistent progress since the 1960s (Mohammadian et al. 2009), city-based approaches to assessing and managing transportation energy and emissions remain at a nascent stage, with a need to harmonize methods similar to the recent establishment of city-scale energy and greenhouse gas (GHG) accounting protocols (e.g., Hillman and Ramaswami 2010, Mehrotra et al. 2011, Kennedy et al. 2009, Ramaswami and Chavez 2013) across cities. With ever-growing data assets and advanced mobility technologies coming to bear quickly (Sarkar and Ward 2016), especially in urban areas (Terrien et al. 2016, Shaheen 2017), there is a unique opportunity to analyze and explore the context-specific responses of cities to these disruptions to shape positive outcomes associated with ACES mobility to the greatest extent possible. Cities are still in the early stages of an evolution toward “big data” analytics, with increased interest in quality data and robust analyses that enable predictive capabilities, allow for higher-fidelity models with finer spatial and temporal resolution, and support new technology and service experimentation (e.g., on shared mobility, automated/electric vehicles, and related infrastructure). By presenting the types of data and modeling platforms emerging across Smart City ecosystems, this paper aims to enable efficient access to the knowledge generated from Smart City peer cities, share knowledge and insights, and benchmark its progress. It also aims to identify gaps in knowledge and practice, which in turn will expose opportunities for the U.S. Department of Energy’s (DOE’s) Systems and Modeling for Accelerated Research in Transportation (SMART) initiative to contribute and gain insight and valuable data from smart city programs.

To fill critical data-modeling-knowledge gaps, improve early understanding of urban mobility transitions and transformations, and identify best practices by cities, cross-city analysis of the seven Smart City finalists served as a critical foundation. This report builds on an initial framework for understanding city data and models (Sperling et al. 2017) for a subset of the seven SCFs (Columbus, Ohio; Denver, Colorado; Austin, Texas; and Portland, Oregon), collecting and organizing data addressing the five dimensions illustrated in Figure 1. Building on that initial study, this report characterizes and benchmarks the current data and model environments related to urban mobility systems as well as Smart City goals and critical mobility indicators with respect to emerging mobility services. The knowledge and information gained from this curation exercise are conveyed in this report. The information is also condensed and presented using template profiles for each city’s data and model infrastructure in the appendix. These templates benchmark existing model maturity and data assets across the seven SCFs in a compact form to enable cross-city comparison. On a broader scale, these seven cities, as a result of being successful SCFs, are representative of the most advanced state of practice and analysis capabilities with respect to the use and uptake of urban data and modeling at the junction of mobility and energy in the United States. 

Figure 1. Initial framework for curating the evolving data and model environments with Smart Cities

In addition to providing a cross-city perspective and reference to the landscape of a city’s capacity with respect to data and models, the initial findings revealed gaps in knowledge and fundamental modeling and analysis capability. Chief among those findings (as further elaborated in the Results section) suggest that few (if any) regional travel models have the capability or the base data to reflect the real-world impact of emerging services such as transportation network companies (TNCs). This is especially critical since TNCs represent the first wave of adoption of “mobility as a service,” which may be accelerated by vehicle automation and digital connectivity technologies (as they mature). The ability to assess and then model “mobility as a service” is an immediate challenge to further clarify the shorter to longer term impacts of ACES vehicle technologies. This and other identified gaps in data and models are noted in the Discussion section.

Case Studies

Initial case studies of the seven SCFs from the U.S. DOT’s Smart City challenge—Columbus, Pittsburgh, Denver, Austin, Portland, San Francisco, and Kansas City—were developed to characterize the existing data and model assets.

This serves as a foundation to explore the extent to which urban mobility data and models are evolving in response to Smart City action planning and with respect to the anticipated disruptions from ACES mobility technologies. The interdisciplinary methods to gather information and elicit feedback from the seven SCFs included:

  • Interdisciplinary data and modeling workshops
  • Mine existing literature and web resources
  • Interviews with Smart City finalists.

These are described in more detail below, followed by results as the summation of these methods.

Data Sources

Three types of data assets were reviewed to extract mobility data, including:

  • Smart City websites – With the Smart City Challenge and the movement it fueled, Smart City websites that relay the city’s goals and objectives and, in many cases an outline of a program to accomplish them, have become prevalent. Some even incorporate a datasharing capacity.
  • Regional web data portals for open-source data – Although the Smart City Challenge has fueled data-sharing and open data systems, many urban data-sharing initiatives pre-date the Smart Cities movement. Many regional entities (often the city, the MPO for the region, or a university partner) publish substantial urban data sets, inventories, and even some application programming interfaces (APIs) for the public and third parties to access. 
  • City-centric mobility research publications – Recognizing that existing mobility information is lacking with respect to new mobility technologies, many public and private actors have proactively performed data collection, analysis, and research to respond to the challenge of improving understanding of current trends. Review of this literature revealed significant ongoing experimentation, data collection, and analytics for specific urban areas (yet with limited emphasis on energy efficient mobility).

Pittsburgh Transportation Data Infrastructure

A key initiative that came to fruition from Pittsburgh’s Smart City proposal is the “data utility,” building on major efforts by the University of Pittsburgh, Allegheny County Port Authority, and the City bringing together over 250 data sets now available at Western Pennsylvania Regional Data Center (WPRDC). The WPRDC was initiated to bring public information to a common platform, making it easy to find and use. Data sets hosted in the WPRDC are provided by the City of Pittsburgh, Allegheny County, Pittsburgh Parking Authority, and Bike Pittsburgh and include bicycle lane use counts and parking utilization. Initial real-time parking data are available for city parking garages. A tripling in parking revenues between 2011 and 2015 (from $5.5 million in 2011 to $17.1 million in 2015) resulted from shifting from old meters and rates to new rates and “pay-by-plate” parking meter technology (paid parking is referenced to the vehicle license plate instead of the parking stall). 

For measuring transportation energy, the City emphasizes use of its energy inventory efforts that include VMT (with odometer readings at city and county levels shared with CMU through a nondisclosure agreement with the Pennsylvania DOT) along with traditional national- and state-level estimates derived from VMT estimates through the U.S. DOT Highway Performance Monitoring System. As to other potential data contributing to the measurement of fuel use, the City and local researchers have not used fuel sales data or vehicle registration data; however, they do have an interest in continuing survey work with the state DOT to better understand shifting fleet composition. 

In terms of impacts of TNCs, the City is working towards a data-sharing agreement with Uber and advancing its own survey work through the City’s Chief Data Officer. At the time of the interviews, the City of Pittsburgh has asked Uber and other TNCs for trip origins and destinations, trip times, time of day, number of occupants, route taken, average speeds, and incidents of crashes. The City is currently exploring moving forward with a subsample of roughly 3,000 Uber drivers and data for the past two years, with initial analyses focusing on use cases for particular areas of the city for benchmarking transit impacts and first and last mile connectivity. 

The results of a 2015 Make My Trip Count commuter survey offer a glance at the complexity of the Pittsburgh commuter patterns (Figure 6). The survey asked where commuters travel from and to, and which modes they regularly use. Compiled from

20,710 responses, the results reveal that less than half of respondents drive to work alone and that many commuters use multiple modes.  

The City of Pittsburgh also actively engages multiple members of the Traffic21 Research

Institute and the Technologies for Safe and Efficient Transportation University Transportation

Center, hosted at CMU. One new data analytics effort housed at the CMU Mobility Data Analytics Center focuses on transportation network modeling, emerging data from intelligent transportation systems (ITS), urban systems integration, transportation economics and traveler choices, and the various sources of data for multimodal transportation systems in Pittsburgh. A significant emphasis of the data being collected focuses on a regional Smart Mobility challenge on how innovative technology can improve mobility. CMU also has a parking data effort funded by the National Science Foundation’s cyber-physical systems program, titled “Matching Parking Supply to Travel Demand Towards Sustainability: A Cyber Physical Social System for Sensing Driven Parking.”

Pittsburgh Transportation Modeling Capacity

The Southern Pennsylvania Commission (SPC) oversees the long-term transportation planning and the travel demand modeling for the region. The SPC TDM is a four-step TDM belonging to the class of “legacy/traditional” travel models, and not the advanced “activity-based” travel models. Traditional travel models consider a person’s travel itinerary as individual, disconnected trips without explicit consideration of the characteristics of the trip maker. The SPC model provides aggregated trip matrices to a static-assignment program, which simulates the trips on the network. The demand generation and network assignment processes happen in a sequential fashion, with feedback iteration across the entire peak period, hindering the model’s ability to model real-time traveler behavior. The SPC model is built using CityLab’s TP+ software and covers a geography of ~1,200 traffic analysis zones. The model considers auto, transit, walk, bike, and freight modes and is currently being run to inform various scenarios related to infrastructure, land use, and economic development of the region; scenarios that are quite traditional with respect to roadway infrastructure planning. The region conducted its last comprehensive travel survey in 2007, and the most recent upgrade to several components in the travel model took place in 2015. The SPC travel model provides the data and analysis that help shape the region’s long-term transportation plans, manage congestion, and meet air quality requirements. Key future scenarios in the MPO’s (as well as the City’s) focus with respect to transportation are vehicle electrification, mode shift and transit use. Key drawbacks of the SPC travel model include 1) lack of advancement in a city-relevant modeling framework (although joint city-MPO modeling efforts are being made on a new bus rapid transit corridor), 2) inability to accurately reflect real-time travel behavior, and 3) lack of consideration to energy impacts.

The SPC covers a large area of Pennsylvania—10 counties, including Allegheny County with Pittsburgh as the county seat. In contrast to other Smart City finalists, the relationship between the City of the Pittsburgh and its MPO (the SPC) is not as tightly coupled as observed in other cities such as Columbus, Denver, Kansas City, and others. In other urban areas, the primary city and its suburbs dominate the planning region with respect to land and population. In Pittsburgh, due to the vast coverage of the SPC and encompassing other population centers within the MPO region, this relationship is not as dominant. The SPC is seen more as a regional partner and a broader metropolitan regional actor. The SPC does publish regional data sets on its websites as an additional source for transportation and other planning data, but it is not considered as city- or urban-specific as the MPOs in other SCFs. (San Francisco is also an exception, due largely to two other major population centers within its MPO planning boundaries.)

[See the appendix for the Pittsburgh Modeling summary template.]

Pittsburgh Takeaways
  • Energy interests are well represented in Pittsburgh’s Smart City vision (the City expressed its desire to have the DOE at the table from the very beginning for the Smart City challenge); the intersections between the information and communication technologies, energy, and transportation themes all came up in the Smart City application. 
  • The Smart City Challenge helped the City to build new partnerships and think about existing partnerships. The intent and purpose of the Smart Pittsburgh consortium were all relevant to the infrastructure decision-makers, including Duquesne Light Company (as the electric utility). The City is also engaged in other activities via the 100 Resilient Cities initiative, in terms of how to build “resilience” into the Smart City proposal.
  • Pittsburgh’s strong partnerships with local universities, namely the University of

Pittsburgh and CMU in the Smart City space, differentiates it from the other SCFs.

Although other SCFs have university collaborations, in Pittsburgh these relations are integral to assisting in both the vision and execution of Smart City goals. 

  • Pittsburgh identified that it was important to define key concerns/goals (energy, mobility, and other) and then scrutinize if the technology offered by vendors matches the need. This was viewed as more efficient than simply receiving technology vendor pitches without a clear mapping into the City’s goals and objectives.
  • Pittsburgh has a unique interest and relationship with TNC service providers, particularly Uber due to its close ties to CMU and being the focal point of Uber’s AV initiative. However, to date, access to TNC data or close collaboration with any TNC company, namely Uber, has not materialized as a result.
  • With respect to travel modeling capabilities, Pittsburgh lags in that the City uses a traditional trip-based travel model lacking behavioral realism. The SPC model (in its current state) lacks the ability to model the mobility/energy impacts of CAV technologies or mobility as a service.             

San Francisco: Smart City Challenge Finalist  

Energy-Efficient Mobility Goals and Metrics (as stated in Sustainable & Smart City Plans) De-carbonization of the region’s electricity supply  o One-hundred percent renewable electricity by 2030Sustainable Travel Choices o Fifty percent Sustainable Trips: aim to reduce solo car trips and make at least half of trips by public transit, ridesharing, biking, or walking Low-energy and emissions “eco-zones” o      Reduce congestion and emissions via low-energy and limited entry zones or roadways (where hazardous levels of emissions and high energy use are reduced) Urban Sensors Deployment o Emphasis on energy, transport, buildings data to be added to open data platform Alternative Fuels o      Electrification and 100% renewable electricity (including large hydro) by 2030 Smart Grids o District energy, microgrids, roadway electrification and electric vehicles Parking Supply and Demand o            Manage and reduce energy use and emissions through SF-Park pilot initiatives
Introduction to the San Francisco Smart City

“In 2012, the SFMTA [San Francisco Municipal Transportation Agency] Strategic Plan targeted a goal of 50% non-driving trips and 50% sustainable trips (walking, transit, bicycling, shared rides) by 2018. The City has achieved a remarkable 50% non-driving mode share 3 years early even while seeing quick growth in its population and employment.” – City of San Francisco Smart City Challenge Proposal

Improving upon shared mobility with CAVs as a community-driven approach was an overarching goal for San Francisco. Reducing energy use, promoting non-single-occupancy driving trips, and expanding city efforts towards CAVs to shape the future of mobility continue to be an emphasis for the City. However, in the more immediate future, the City of San Francisco is concerned about the impacts of the proliferation of the use of TNCs such as Uber and Lyft on its streets, but there is a real lack of comprehensive data to help the public and decision makers determine how best to harness TNCs for reducing congestion, enhancing transit ridership, and attaining other system performance goals. For example, TNC trips are not all new trips to the city, some simply replace otherwise single-occupancy commutes. However, there is insufficient data to determine the percentage of TNC trips that are simple mode substitutions or net new travel.

Another consideration for the City has been accommodating the more than 45,000 TNC drivers operating in San Francisco from a prior system of only approximately 1,800 taxi drivers (Castiglione et al. 2017; San Francisco Municipal Transportation Agency 2016). The region also welcomes 50,000 regional visitors daily, noting that without adequate information on mobility choices, many may choose auto-based modes over transit, thereby adding to congestion, parking, and other related impacts.

In addition to the concern around TNC observability (as noted by Castiglione et al. [2017] and City staff), challenges due to rapid growth leading to lack of affordable housing and longer commutes for lower-income populations (from areas with more limited transport options) have also been noted.

With the surrounding region already seeing AV testing, the San Francisco proposal attempted to extend efforts and synergies among shared mobility, CAVs, and transit in ways that might reduce costs and time for travel, as well as the need to own a vehicle. This approach aligned with prior City goals of a 50% non-driving mode share by 2018, which had already been met by 2015. 

The City of San Francisco not only viewed the Smart City competition as an opportunity to advance transportation partnerships and innovation in its region, it has continued to make significant progress since. While 16 pilot programs were proposed responding to neighborhood, city, and regional transportation challenges, the City has since received an advanced

transportation grant that is funding four pilot programs from its Smart City application; with San Francisco County Transportation Authority (SFCTA) adding in Treasure Island pilot projects that include electronic tolling and automated shuttles. The U.S. DOT invested $11 million in the City to help advance driverless shuttles, ride sharing (as connected carpool and public transit lanes), dedicated curb space, etc., and smart traffic management. In all, since the competition, the City has moved from proposing to develop 16 pilot programs in three years to developing six pilot programs in five years. One major emphasis has also been pursuing a grant from the California Air Resources Board to electrify portions of the TNC fleet. Another major emphasis is to streamline public–private partnerships, as catalyzed by Gillian Gillett, the mayor’s Director of Transportation Policy. 

As to the relationship of departmental roles and responsibilities to energy-efficient mobility systems, the Department of Environment oversees analyzing all the particulate matter and VMT, taking emission factors for fleet averages, and determining how to clean the electric power grid as relevant to emissions. Shipping, bus, and ferry planning, investment, and operations are led by the San Francisco Municipal Transportation Agency (SFMTA). The San Francisco Public Utilities Commission is responsible for the Community Choice aggregation initiative and focuses on power supply/grid side. City energy accounting is also under the Department of Environment (both fuel/electricity together). Pacific Gas and Electric Company and Hetch Hetchy Water and Power are the sources of electricity for city buildings and EV fleets. City staff are also working on an emerging mobility report, addressing mobility providers and their operations, and prioritizing key challenges and alignment of agencies and their respective goals. For example, the SFMTA with SFCTA most often have aligning goals, yet there are times where publicprivate partnerships affect relationships due to different governing structures (e.g., SFCTA has a board of supervisors, while SFMTA has a separate board, with some members appointed by the mayor). 

San Francisco Transportation Data Infrastructure

At the time of its Smart City Challenge submittal, San Francisco had one of the most mature open data programs of the SCFs. This included an open data policy, open data platform, and central clearinghouse (called SFOpenData), which includes over 350 data sets (of which 30 were transportation related). Since then, new data sets have been added as open, machine-readable data, with increased interest in hosting data on shared, connected, and automated vehicles (once relevant policies are in place) for the City to help enable data-driven smart cities and energyefficient mobility entrepreneurship and innovation. Figure 7 shows an example of the TNCs Today Data Explorer – Fridays Summary Statistics.

Figure 7. Sample of the TNCs Today Data Explorer – Fridays Summary Statistics

(Castiglione et al. 2017)

“The success of TNCs in attracting rides in San Francisco and other cities reflects the high unmet demand for premium services and the extensive benefits they provide. Initially TNCs offered some distinct advantages over taxis, including the ability to easily reserve a ride, the ability for both driver and passenger to contact each other and to know the location of the other using GPS, ease of payment, cheaper fares, shorter wait times, and more availability at all times of day due to a larger supply of vehicles. Taxis now offer some of these features, although the supply of taxis is still significantly smaller than TNCs, and taxi fares are higher.”  – SFCTA (2017) 

San Francisco’s OpenData initiative, launched in 2009, continues to support the San Francisco Smart Cities initiatives that aim to meet GHG reduction goals and improve and increase public transportation service. The City is also very focused on TNC data. Access to data on TNCs, from a planning perspective, was identified as a near-term priority challenge and opportunity. Through initial collaborations with Northwestern University, a report titled TNCs Today: A Profile of San Francisco Transportation

Network Company Activity was issued that characterized San Francisco TNC traffic using data skimmed from the API interfaces of TNC companies in November and December of 2016 (SCFTA 2017). The report was complemented by an initial TNC data web platform with data available for download ( and a visualization platform ( for exploring available San Francisco TNC data spatially and temporally. The study estimated that TNCs accounted for 15% of all intra-city

trips, where previously the City estimated TNCs accounted for 7% to 9% of all trips; however, neither estimate can be corroborated from other data sources. 

San Francisco Transportation Modeling Capacity 

The City and SFCTA interface on their larger model for the purposes of analysis of energy and transportation: the SF-CHAMP model produces all the relevant VMT data, and the City Department of the Environment is in charge of the GHG data. Currently VMT estimates for performance measure reporting are taken from the SF-CHAMP model, yet this model does not have vehicle fleet composition data (e.g., percentage of EVs or average fuel economy of internal combustion engine vehicles) as revealed from state registration data bases. On the modeling side, energy/GHG impact assessment is also done using the SF-CHAMP model as the primary city model. There is also a regional Metropolitan Transportation Commission/Association of Bay Area Governments as the MPO model that City staff were less familiar with. An observation from the City was that the regional model was ill-equipped and often lagging in terms of directly helping to understand city energy, VMT, and GHG emissions. This included the identified challenge of having over 45,000 new (Uber/Lyft) vehicles on the city’s streets not being factored into the model. Figure 8 shows TNC activity as part of the modal split for an average weekday.

Figure 8. Profile of TNC activity as part of the modal split (average weekday)

Areas of interest identified by the City of San Francisco for improving the SF-CHAMP model capacity included:

  • Learning what has been done or can be done to make the models accurately reflect new modes of transportation that were not foreseen when the models were built 
  • Taking advantage of new technology (sensors, “big data,” machine learning for validated predictive analytics) to build entirely new models that go beyond anything developed so far
  • A better understanding of the “why?” of data collection and travel demand modeling,

e.g., the types of questions models are currently unable to answer, the problems and limitations as well as the questions that—in theory—could be answered by implementing new data models.

[See the appendix for the San Francisco Modeling summary template.]

San Francisco Takeaways


  • SFCTA is the main agency for city-regional modeling on energy and transportation, where VMT data are extracted by the San Francisco Department of Environment for cityscale energy and GHG accounting. However, the current model is limited with respect to the impact of TNCs as well as EV inventory.
  • There is a concern over the approximately 50,000 new vehicles on city streets from TNC activity that are not being factored into the model. The model is ill-equipped to understand the VMT and GHG emissions of TNC vehicles.
  • Fast changes are underway, with multimodal surveys indicating that TNCs went up from 2% in previous years to an estimate of 7% in 2017. 
  • Remaining questions for the City: TNCs are likely taking some share of public transit— but how much? The TNC vehicles are likely also taking some share from private cars— but how much? Understanding these questions (and having available data on this) is critical to reducing congestion and pollution and shaping a TNC system that reduces private car trips.

An opportunity was identified to have more targeted data collection with trips to and from the airport (as a City-owned asset). Today, San Francisco Airport charges $3.50 for every TNC trip, generating upwards of $30 million in revenue per year. Methods to collect financial data across cities via public disclosure reports can provide an early critical understanding of TNC trends/impacts.

Energy and Mobility Goals, Model Capacities, and Data-Driven Performance Indicators by City

With respect to the top-level motivations and overall objectives, although the SCFs share some common themes and objectives, at the top level, Smart City priorities and motivations are unique and distinct to each city. The common element across all cities is the need to bring data collection, management, analysis, visualization, and modeling to the core of the Smart City operations and decision-making process. Data-driven decisions, data-informed policy, and objective data-validated performance measures are all strong elements of each city’s platform, both in the mobility/energy space as well as water resources, pollution, air quality, equity, health care and other areas vital to the quality of life of its citizens. While mobility plays a large role in the Smart City objectives, it remains balanced with (or seen as a key enabler of) other concerns of health, equity, economic vitality, and jobs. Mobility is unique in that it is seen as a tool to address concerns over a wide variety of issues. This is best illustrated in the Columbus Smart City portfolio where some of the mobility-based pilot programs are undertaken with the primary objectives of improving healthcare, access to jobs, and improve overall sustainability of the city, particularly for the underserved communities such as the Linden neighborhood. In short, whereas mobility is seen as a cornerstone to Smart City initiatives, at the motivational and objective level, simply “improving mobility” is rarely directly stated, but rather it is a derived objective, or a means to an end of the more fundamental needs of the citizens in the community. 

Likewise, the data movement within cities is not limited to transportation or energy in transportation, but, similar to base objectives and motivations of Smart Cities, transportation data, metrics, and analysis play a substantial role in assessing the quality of several motivational concerns such as access to health care, employment, basic resources such as food and services, social equity, and the ability to support an economically vibrant community. As an opportunity to fill an identified gap, a “quality of mobility” metric is needed—not one that is based on vehicle speed or number of transit routes, but one that reflects the ability of the transportation system to connect inhabitants with the basic goods, services, and employment which define a quality urban environment, or as was stated concerning the Smart City Challenge: “Transportation is not just about concrete and steel. It’s about how people want to live.”  Engagement with any city requires knowledge of the elements that shape its primary motivators in the Smart City arena, and to then identify how mobility and energy goals may align. Although there are commonalities between different cities, each is unique in its objectives, shaped by the current economic condition, history of the city, and specific geographic/climate-based context. For example:

  • The Columbus Smart City program is deeply rooted in supporting a continually growing and vibrant economy that will provide benefit to all its citizens, with specific emphasis on how to address the underserved communities effectively as the city grows. Having experienced going from the world capital of buggy production in the early 1900s to a nearly  extinct industry in less than ten years, Columbus is determined to capitalize on the emerging paradigm shifts in mobility, harnessing it as a growth engine, and not be left behind.
  • Portland, with a history of air quality issues stemming from its industrial heritage, keeps emissions and air quality at the forefront of its objectives in its Smart City objectives, emphasizing mobility options that not only connect its citizens but provide a long-term air quality / emission benefit while complementing its existing transit infrastructure.
  • Austin, like Columbus and most of the SCFs, is growing rapidly. The potential to outgrow its existing road and transit capacity is seen as a threat to continued economic prosperity. Austin is experimenting with emerging technologies to not only decrease energy use (EVs), but equally to keep congestion at bay to sustain the growth of the city.
  • Denver, having invested in light-rail and its expansion (most recently to the airport) is emerging as the focal point city of the Rocky Mountain Front Range population centers. Due to Taxpayer’s Bill of Rights (TABOR) legislation as well as the state Transportation Commission policy, any significant roadway capacity infrastructure investment is likely to be tolled-facilities moving forward. As the Denver metropolitan region continues to grow and develop, methods to leverage emerging ACES technology, in concert with its newly enhanced light rail system, are at the forefront to ease congestion, and also to continue to improve air quality and sustainability.
  • San Francisco with its Bay Area Rapid Transit (BART) and street cars, is motivated by its sustainability goals as part of the larger California state initiative. As the originator of the TNC movement, the growth of TNC service and its secondary issues (street and curbside congestions), which could potentially threaten progress toward its climate sustainability goals, are at the top of their agenda.
  • Pittsburgh, a legacy industrial city in rugged western Pennsylvania, is experiencing growth after decades of decline and is faced with how to best leverage its aging infrastructure assets in combination with ACES technology to capitalize on growth opportunities as fiscally efficiently as possible. 
  • The Kansas City metro region, located in two states and straddling the Missouri River, possesses few geographical impediments to continued urban sprawl in any direction, yet core city growth continues. Technology options to encourage continued, efficient urban development and economic growth is at the core of the region’s motives to invest in Smart City technology.

Although the motivating factors are varied, each of the SCFs is rapidly experimenting in the emerging mobility options and associated data space as funding resources allow.

Data Infrastructure

Collaborative data sharing initiatives are a common theme across all cities. These approaches range from establishing the Smart Columbus Operating System (SCOS) and “data sandbox” in Columbus, “data utility” in Pittsburgh, the data PORTAL in Portland, an enterprise data management system in Denver, a “one data system” and “data rodeo” in Austin, an awardwinning “XAQT” platform in Kansas City, to “DataSF” in San Francisco. Existing data sharing and exchange within the cities are currently enabled by collaborative efforts in and among the city, surrounding jurisdictions, and transit agencies, with primary partnerships with the local and regional entities such as the MPOs, universities, philanthropic institutions, and economic development interests. In most instances, these organic data-sharing efforts pre-date the Smart City Challenge with regional data warehouses and geographic information system data set sharing from the city, MPO, and/or university partner. However, these initiatives share a common goal of supporting a sophisticated, dynamic data exchange capacity (not just warehousing and sharing) with the goal of enabling API-based data access to third parties. This dynamic data exchange capacity is at the heart of the Columbus SCOS development and is also prominent in the objectives in all the other SCFs. However there exists no clear common path across all these cities to obtaining this dynamic information exchange, nor are there standards, nor a common framework. Each city approaches this independently based on the resources and expertise within its data infrastructure partner collaborative, which is substantial in many cases. 

The development of city data infrastructure for Smart Mobility is paralleled by a data infrastructure initiative by the U.S. DOT to create a Roadway Transportation Data Business Plan for state DOTs and local agencies (U.S. DOT 2013). The Federal Highway Administration Office of Operations began this effort in 2013 to help state DOTs and local agency staff charged with mobility data-related responsibilities, develop, implement, and maintain a tailored plan for “Roadway Travel Mobility Data.” This was in response to increasing data reporting requirements from MAP-21, the Moving Ahead for Progress in the 21st Century Act (P.L. 112-141), and the increasing amount of base data being collected by each agency as part of its ITS programs. Base data collection technologies for traffic and road infrastructure are undergoing a technological revolution such that vast amounts of disaggregated electronic data from sources such as industry probe data, imagery, and new sensors are created, which need to be controlled for quality, aggregated, archived, and reduced to summary information. The goals, motivating pressures, and framework encompassed in this Federal Highway Administration effort to assist DOTs are analogous to the current need within Smart Cities to create dynamic data/information exchanges. While the DOT effort is motivated by making data-driven decisions to maximize the effectiveness of limited federal and state highway funds for developing, maintaining, and operating highway transportation infrastructure, the data initiative for cities is motivated by the need to advise and guide investments of limited resources at the local level to the greatest effectiveness for Smart City objectives and provide continuous feedback on the performance of those investments. The U.S. DOT Roadway Transportation Data Business Plan addresses both technical (integration, sources, tools, and technology) as well as institutional (costs, roles and responsibilities, and data governance) aspects. 

The SMLC’s data expertise and subject matter expertise within the energy spectrum provide a basis to be a major contributor to the Smart Cities data infrastructure program. These new data exchange platforms that rely on agile development, distributed computing, federated data assets, and sophisticated rights tracking are all areas that the DOE Laboratories have had to navigate in their data systems (for example, the Transportation Secure Data Center, Alternative Fuels Data Center, Fleet DNA, and others). The inclusion of energy and emissions data and metrics, particularly as they intersect the mobility network, is challenging for nearly all cities. This  provides the DOE with an opportunity to influence and encourage the integration of energy metrics into urban mobility performance measures and inform the collection of key data.

As the data infrastructure and exchanges within cities continue to evolve and mature, standardized APIs across a variety of core data areas would enable outside parties to efficiently access and leverage public data sources for analysis/visualization/data products. For example, as DOE analysis programs mature in this space (such as the Electric Vehicle Infrastructure Projection [EVI-Pro] model to estimate the number, type, and location of electric vehicle supply equipment to support EV adoption goals), developing protocols in these tools to operate off emerging Smart City data platforms will accelerate the data exchange infrastructure and provide motivation to standardize protocols and APIs. Until standards emerge, transportation and mobility data at the urban level will continue to be ad hoc, enabled by collaborative efforts within cities, and will appear somewhat chaotic when viewed at the national level. Partnerships and collaboration with local entities (universities, non-profits, business community, MPOs, Clean Cities coalitions, transit agencies, and other jurisdictions) will remain a viable path to open data exchange enabling research and data-driven Smart Cities. 

These emerging open data exchanges at the city level would benefit from new data collection methodologies, techniques, and sources. For example, speed data collected through road-side sensors were the norm for state DOTs until about ten years ago. The cost and ability to deploy and maintain traffic sensors limited its wide-spread proliferation (particularly on lower road classes such as minor arterials and local streets). About ten years ago, data from vehicles selfreporting their location and speed based on onboard global positioning system equipment as well as personal data from smart phone travel applications began revolutionizing traffic speed and travel time reporting. Thanks to these technological advancements, traffic conditions are known with confidence across the entire network (and not just a handful of locations with installed sensors). These same data are also being mined to identify incidents, predict travel times, and provide performance metrics across the national highway system. Such data are the basis for the National Performance Management Research Data Set for the reporting requirements in the most recent MAP-21 legislation. 

Similarly, new technology is providing cities with data sets for arterial roadway performance, bike and pedestrian activity, infrastructure inventory and utilization, parking operations, and many other areas at price points, performance, and scalability that far exceed traditional data collection methods. As these capabilities come on line, the Smart City networks benefit from communicating best practices and lessons learned from early adopters to other Smart Cities (while communicating the data through the data exchanges) to quickly leverage the benefits of these advances. Similarly, the emerging Smart City data exchanges provide opportunities to the SMLC to access new data sources critical for SMART research tasks, which in turn contribute to the progress of Smart Cities. Many of these new technology service-driven initiatives create “Big Data” sets that DOE laboratories can navigate and explore more readily (through their high- performance computing capabilities) than the computing resources in a typical mid-size city. The procurement of statewide trip data by the Ohio DOT through an industry source (INRIX) is one such example. The beginning and ending locations and intervening waypoints (between one second and one minute apart) of approximately 1% to 2% of all trips in Ohio represent tremendous opportunities for cities to understand existing travel patterns for numerous applications. However, the data management and processing to deal with giga-byte data sets are not yet readily prevalent in Columbus or other Ohio jurisdictions. 

One last shared theme that emerged among many (though not all) SCFs was the need for better

TNC data to understand the extent of market penetration (what percent of travel is being taken by

TNCs), impacts to congestion, influence on car ownership, propensity to use transit, contribution

(positive or negative) to a city’s sustainability goals with respect to VMT, impact on land use (parking and curbside requirements for new developments), and impacts on revenue (due to parking demand declines or increases due to access fees). Although many cities are posing such questions about TNCs, San Francisco stands at the forefront with respect to its concerns about the influence of TNCs to its overall sustainability goals, citizen mobility, and energy footprint. However, the dominant TNC companies, Uber and Lyft, are inhibited from sharing data over concerns of commercial competitiveness. With the critical shortage of base data, data collection and analysis centered about major inter-modal mobility hubs, such as airports, that collect access fees (and thus data) for TNC as well as other ground modes (parking, car rental, car-sharing) offer a prime opportunity to observe mode choice and travel behavior shifts. Such data collection and analysis activities were undertaken within the DOE EEMS initiative in 2018 as part of the overall effort to address the larger research question of whether mobility-as-a-service (with TNCs being the leading edge) will have a positive or negative energy impact, as well as the factors that impact this determination.

Establishing methods to estimate key impact parameters from available data is critical for Smart Cities to monitor performance as well as augment data assets for modeling long-term impacts of ACES technology in the energy and mobility domains. 

Modeling Capacity

Since the Federal-Aid Highway Act of 1962, standard planning processes managed by MPOs have been required in the United States for urbanized areas with a population greater than 50,000. As a result, a large majority of the travel modeling practice (and subsequent expertise) is concentrated at the MPOs that serve the major cities in the United States, with a few notable exceptions. For the SCFs, this pattern holds true for all cities except San Francisco where three large population centers in the Bay Area (San Francisco, Oakland, and San Jose) have prompted the county of San Francisco to maintain a separate transportation model specific to San Francisco interests. Typically, TDMs developed and operated by the MPOs inform the long-range transportation plans for the region. These models are also used on a regular basis to study and estimate the impact of proposed infrastructural improvements in a city/region, such as adding a new lane to an existing roadway. Over the years, more and more have been required from these models. Initially planned for highway capacity concerns, their scope has been increased over the decade for emissions (planning air quality), transit and transit-oriented development, and ITS operations, and now they are being called upon for Smart City operations and assisting with exploring the anticipated influences brought upon by emerging ACES technologies. 

As with previous challenges to transportation modeling, the ability for TDMs to reflect the impact of emerging ACES technologies lags in capability but is quickly evolving. Over the past decade and a half, TDMs have evolved more sophisticated methods to reflect travel behavior choices at the individual traveler level. This new methodology, referred to as ABM, models the decision process of each member of a representative population for all activities through a typical day, then combines all the individually generated trips into an overall aggregate representation of travel demand, referred as the trip table. Previous methods (referred to herein as a four-step approach) use more homogenous assumptions about the population and its travel needs to directly estimate a trip table, concentrating mainly on journey to work periods (and other peak periods) of the population. Similarly, the assignment of these trips onto the network model (encompassing roadways, transit, pedestrian, cycling, etc.) has also evolved. Modern methodologies, referred to as DTAs, assign the trips to the network in small time increments (such as 5-minute, 15-minute or 1-hour intervals) rather than the general peak hour assignment of older methods (referred to as static assignment). While the previous generation of TDMs had the travel demand and network assignment components work in a sequential fashion (demand is first generated for the peak period and supplied to the assignment module for simulation on the network), the more advanced methods currently being adopted by the MPOs have the demand (ABM) and network assignment (DTA) components communicate on a more regular basis. The ABMó DTA interaction at regular intervals allows for accommodating realistic representation of real-time travel behavior as well as responses to recurring and non-recurring network congestion. For instance, as the travel time increases in response to congestion, trips may be assigned to different routes. If the DTA and ABM are tightly coupled with feedback loops, congestion or incidents on the road network may prompt a traveler to change the mode to transit and thus alter the trip table. 

The TDM capacity and capability within the SCFs range from the most modern methodology implemented (an ABM with a DTA) to more traditional approaches (four-step with a static assignment). The pace at which an urban area adopts the newer methodology varies based on needs, resources, and the cyclic nature of model development, which tends to be on an 8- to 10- year cycle. The ABM with DTA framework is the most modern and is anticipated to be sufficient to support ACES modeling. However, as revealed by the interaction with the SCFs, as well as supported by the feedback from the modeling and data workshops, current TDMs do not have the capacity to provide  information to all (or even some) of its research questions with respect to emerging ACES technology, and rightfully so. This is not surprising as modeling is generally reactive to needs. Without sufficient data upon which to base the behavioral and traffic expectations, any attempt to build predictive capacity into a TDM would not yield results with any level of confidence. Although the current TDM frameworks (ABM and DTA) are anticipated to be sufficient to reflect the impact of ACES technology, base-level knowledge to build the human and traffic behavioral responses remains lacking. 

As an example, if a “taxi mode” were to be incorporated in a TDM, it would have facilitated (or begun to facilitate) the modeling of TNCs in cities. However, none of the models reviewed has a “taxi” option in them because the market share of this mode has historically been negligible (below 2%) in most cities. The prevalence of taxi trips in cities was so small with respect to other, more-dominant modes that no need was apparent for inclusion of taxis in TDMs. Roughly five years ago, expectations that TNCs would gain rapid market share, much less thinking about incorporating TNCs as a new modal option, was not a consideration. Now that TNC trips are increasing in number, many cities are considering including this as a separate mode in their mode choice models. However, the available base data to understand consumer behavior is currently insufficient to build the models. The case for building capabilities to model AVs will be similar, as it is not possible to understand the impact of AVs on travel until people experience them and their observations and preferences are recorded.

A cross-city comparison of the model capability is provided in Table 2. A review of model maturity across different cities reveals that many cities are building up their modeling capabilities, be it moving from trip-based modeling approaches to more advanced (and behaviorally realistic) ABMs or incorporating DTA to depict real-time travel behavior more accurately across several modes. As evidenced in the table, no two TDM profiles are the same, with each model being customized to the needs of a city/region. 

  • Of the cities fully characterized in Table 2, four incorporated ABMs while the other three rely on the four-step method. (Note that even within ABMs there are different levels of fidelity, but such characterization is beyond the scope of this review.)
  • Three of the cities use DTAs, while three use a more traditional static assignment approach. The Austin official model is static, though it has been extended by TTI to a DTA.
  • The dates of the existing and planned upgrades (when known) indicate a period for model development ranging from 5, 7, and 13 years for Columbus, San Francisco, and Denver, respectively.
  • As noted, none of the models reviewed has a specific TNC mode (or even a taxi mode) incorporated in its mode choice models.
  • Some models had a multitude of sub-models to capture travel from special generators (airports, universities, etc.), while others had limited sub-models (and thus run more rapidly.) 
  • Scenarios of interest varied. While some focused on developing and testing scenarios related to technology, others focused on changes in land use patterns in the short- and long-term futures. 
  • Model capacity and capability were estimated based on review of the framework (ABM/DTA) and level of coupling. Such an approach provides a common platform/rubric to initiate a dialogue among cities.

The TDM model characterizations in Table 2 are for the most commonly used model in each city. It is not unusual for different models, or even model variants, to be active within a city, particularly in cities where universities are active in urban modeling research. Austin, for example, with the University of Texas at Austin and Texas A&M’s TTI nearby, has been the subject of various research experimentations leveraging various aspects of the TDM developed by the local MPO (CAMPO). A 2017 research study by the University of Texas at Austin (Liu et al. 2017) leveraged the base CAMPO model and supporting data using open-source ABM software, MATSIM, to explore the anticipated effect of fully automated demand responsive service (“Automated Uber”) based on socio-economic behavioral estimations derived from a literature review. Likewise, TTI augmented the CAMPO model with better traffic response (in essence, a DTA) to assess the impact of consumers’ change in vehicle routing in response to external (smart phone app) stimuli. 

The discussion of the Austin CAMPO model touches on another theme and identified a research gap emerging from the data and model curation task. Model capability is also correlated to model complexity. As TDM models progress from four-step to ABM and from static to dynamic assignment, the data, computing, and personnel resources to build, maintain, calibrate, and exercise the model also escalate. As a result, the choice to advance the model from traditional to a more modern framework and methodology is also a fiscal commitment that falls on the urban area. The gap or research question that emerged from the model capability assessment exercise is whether cities with a traditional framework (four-step and static assignment) can use or leverage their existing model to elicit the impacts of ACES technologies or whether a new model development investment required. In other words, can more traditional (less expensive) modeling tools be leveraged with less fiscal investment to estimate the impacts of ACES and plan appropriately for the urban area? 

As a corollary to the observation above, the modeling and analysis needs of the primary city within the region typically differed from those of other less populous and less-dense jurisdictions. TDMs originated and continue to evolve from a regional metropolitan trip context, emphasizing journey-to-work (peak hour) concerns that impact the volume to capacity ratios on the principal highway transportation network. While peak-hour highway congestion is a problem for jurisdictions across the board, new mobility technologies present an additional layer of complexity in the context of central city and other jurisdictions with significant population and job densities. This is experimental ground for new mobility technologies that unclog city streets, address parking availability and efficiency, and link commuters the last mile between employment and other opportunities with regional transit and other mobility-as-a-service offerings. Such areas are struggling with a host of complex mobility issues induced by density that most other suburban localities are not. Although existing TDMs acknowledge a portion of this diversity by using varying sizes of Traffic Analysis Zones, current modeling frameworks do not accommodate the need for high density areas to model the transportation network in corresponding “micro-resolution” needed to reflect the impact of ACES technology such as ridehailing, shared scooters, shared bicycles, and the curb congestion and parking impacts that they induce (particularly in the context of identifying terminal access/egress times and simulating them on the network). The central cities and some sub-jurisdictions in poly-centric cities require such capacity whereas the greater metro region does not. This points to a need for TDM frameworks to provide a “multi-resolution” approach, adapting higher resolution in the mobility network where and as needed. 

Not listed in Table 2 are the types of energy and emissions outputs from the models. The TDMs were all similar in that they incorporated the EPA MOVES model, not unsurprising since airquality analysis is mandated in the transportation planning processes. Whereas the modeling by MPOs was consistent with federal planning requirements, additional energy analysis was typically conducted at the jurisdiction-specific level. Motivation for doing so was either organic to the city’s goals (such as Denver) or mandated at the state level (such as San Francisco). Columbus is pursuing similar objectives to provide performance metrics in the reduction of petroleum-based emissions as part of the electrification program. Although there may be some similarity in approaches due either to prevailing literature or state regulations, there is no common methodology widely linked to TDMs for estimation of energy usage. (Note, California regulations enforce uniformity for California jurisdictions.) The EPA MOVES model, which provides link-specific emissions, remains the current standard. Evolving methods that reflect the mixture of the actual consumer fleet of vehicles (based on vehicle registration records) and project alternative fuel vehicle adoption rates remains a gap in TDM methodology. 

Table 2. Cross-City Model Capability Comparison Matrix
Cross-City Comparison and Summary of Model DetailsColumbusPortlandPittsburghAustinSan FranciscoDenverKansas City
4-Step (4S)/Advanced 4-Step (A4S)/Activity-Based (AB)ABAB4SA4SABAB4S
Static Assignment (SA)  Dynamic Assignment (DA)DADASASA/DADASASA
Last Upgraded2004201020152010201220102015
Next Upgrade2017??????20172017??

TNC Mode Included? (Y/N)                   N                     N                     N                     N                     N                     N                       N

Special Generator      A– Airport, F– Freight,      IE – Internal/External Trips      U – University, O – OtherF, IEA, F, IEA, IEA, F, IE, U, O (Hospital, Prison)A, F, IEA, F, IE, U, O (Mountain / Casino)A, IE
Scenarios  Considered/Tested     I – Infrastructure        
   D – Demographic    L – Land Use    EN – Energy    EC – Economy    T – TechnologyI, D, TI, D, L, EN, EC, TI, L, EN, ECI, D, L, ECI, L, EN, EC, TI, L, EN, ECI, L, EN, EC 
Model Capacity/Capability Level     A – Advanced    H – High    M – Medium    L – LowHHLMHML


This review assessed the dynamically evolving state of urban mobility data and models, along with Smart City goals and priorities in the mobility and energy space. City data infrastructure and mobility modeling capabilities were characterized for the seven U.S. DOT SCFs according to their ability to support ongoing evolutions with emerging mobility technology related to vehicle automation, connectivity, electrification/efficiency, and sharing, frequently referred to as ACES. This baseline assessment of Smart City data and modeling capacity was created to explore how these systems, emerging services, and analysis capabilities are evolving, so to best identify gaps in data, practices, modeling and analysis, and to map any gaps to collaboration opportunities with the ongoing DOE research initiatives in the EEMS program. This research initiative is part of a portfolio of research projects pertaining to Urban Science whose overall objectives include: 

  • Providing cross-city and high value data sets that are more widely available across networks of cities, private and public mobility providers, and research communities
  • Harnessing the power of emerging city data and observability to enable upgraded modeling platforms to help visualize, conduct analyses, and advance urban mobility model environments as these systems transition and transform in response to multiple disruptive changes that impact energy productivity of mobility, urban infrastructure systems (e.g., physical, natural, societal, and cyber) as well as associated travel behaviors and choices.
  • Harmonizing approaches in data and modeling by developing common methods to observe transitions in urban mobility and the energy impacts from emerging mobility technology 
  • Addressing specific knowledge and data gaps as critical early-stage research. Exploring the impacts of, preparing for, and shaping transitions in cities at the intersection of mobility and energy is particularly needed (to inform performance in enhancing mobility energy productivity, infrastructure modernization, revenue diversification, and choices).

The knowledge and information gained from this assessment of smart city data and models are conveyed in the preceding detailed report. On a broad scale, these seven cities, as a result of being successful SCFs, are representative of the most progressive and proactive jurisdictions with respect to informing responses to emerging mobility technology, as well as balancing that with energy and emissions concerns. The lessons learned, take-aways, data and modeling gaps as well as opportunities for DOE collaboration are summarized below into three sections corresponding to upgrading data infrastructure, modeling and analysis capability, and overall Smart City gaps and opportunities.

With respect to Smart City data initiatives …

Each SCF prioritizes a robust and continuously upgraded data infrastructure to monitor and inform decisions and to provide performance-based measures to assess progress toward its goals. 

  • The need for robust data sharing and exchange platforms is a common theme and initiative within all seven SCFs. Most cities are pursuing this through collaborative efforts led by local partners (universities, MPOs, non-profits). As a reference example of implementation, the U.S. DOT Smart City awardee, Columbus, Ohio, is developing the Smart City Operating System as the core enabler of its portfolio of projects in the transportation and energy space moving forward. 
  • Most data infrastructure efforts are based on data warehouse and/or legacy geographical information system architectures. Such approaches offer the capacity to store data sets but are less adept at real-time transaction interfaces such as APIs. Legacy approaches are also challenged by complex user access rights needed to protect personal privacy as well as navigate licensing of commercial data sets. Newer approaches based on modern internet and smartphone application infrastructures involving agile programming are more adept at enabling robust data sharing.
  • Most urban data initiatives incorporate existing mobility data gathered using legacy approaches, such as deployed sensors or data from existing services such as public transit or parking revenues. New technology-driven, commercial, crowd-sourced, and/or internet-of-things–based methods have the potential to scale quickly; minimize cost; and provide timely, even real-time, data availability, but require big-data expertise and computing resources typically beyond that of most municipalities. This presents an opportunity to leverage core DOE data expertise while gaining access to modern, relevant Smart City research data.

With respect to Smart City mobility modeling and analysis capacity …

The TDM capacity and capability within the SCFs range from the most modern methodologies such an ABM coupled with a DTA network model to more traditional approaches (four-step with static traffic assignment). 

  • The pace at which an urban area adopts the newer modeling methodologies varies based on needs, resources, and the cyclic nature of model development, which tends to be on an 8- to 10- year cycle. The ABM with DTA framework is the most modern and is anticipated to be sufficient to support most cases of ACES modeling. However, as revealed by the interaction with the SCFs and supported by the feedback from the modeling and data workshops, current TDMs do not have the capacity to inform cities on emerging ACES mobility technology due mainly to a lack of research data on emerging modes. This is not surprising as modeling is generally reactive to needs. Without sufficient data upon which to base the behavioral and traffic expectations, any attempt to build predictive capacity into a TDM would not yield results with any level of confidence. This is the situation that TDM finds itself in at the present time. 
  • TDM modeling capacity within the seven DOT SCFs is typically housed within the corresponding MPOs, with some exceptions particularly when there are multiple principal cities within the MPO’s region. At present, for example, Columbus, has adopted modern TDM methodologies and has implemented state-of-the-art models with the latest in ABM and DTA, while Austin relies on a more traditional four-step approach. In either case, cities see the TDM primarily as a rearward-facing tool, informing traditional

mobility (vehicle and roadway based) and not dynamic enough to inform on quickly emerging mobility technologies such as deployment of automated shuttles or management of TNCs such as Uber and Lyft. Tools to address the latter are in critical demand. Current TDM practice is resource intensive, requires large amounts of data, and takes months if not years to construct and calibrate. Tool suites that are nimble and agile are required to assist in the decision space surrounding the rapidly evolving urban mobility landscape and would be a welcome advancement.

  • The standard outputs from TDMs related to energy are aligned with the EPA MOVES module that provides emissions estimates based on the operating speed and volume of roadway segments, which can be adapted for viable energy estimates. More sophisticated energy tools that integrate with TDMs are needed to tailor energy estimates based on consumer fleet composition as revealed by vehicle registration data, amount of shared ridership, and projections of future vehicle mix and ridesharing. This will help align TDM practice with future energy-efficient mobility systems priorities.
  • The needs of the primary city within an MPO region with respect to emerging mobility cumulate on top of that of less-populous, less-dense jurisdictions. The denser central city has become the experimental ground for new mobility technologies due to the concentration of people and activity and exhaustion of road and parking capacity. Traditional TDMs, although having varying size traffic analysis zones, do not accommodate the need for “micro-resolution” within dense urban development where parking availability and curb space congestion may govern mode choice. This points to a need for TDM frameworks to provide a “multi-resolution” approach, adapting higher resolution in the mobility network where and as needed as population and activity densities dictate.

With respect to overall Smart City gaps and opportunities at the junction of mobility and energy …

Aspiring Smart Cities everywhere are seeking to harness new data, communication, and mobility technologies for the benefits of its citizens. Mobility is unique in that it is not perceived as an end goal or objective in itself, but a means to an end for improved economic productivity, equitable access to health care and employment for citizens, and as an overall enabler to a higher quality of life. 

  • The impacts of TNCs such as Uber and Lyft are within the spectrum of awareness for most cities with respect to the benefits provided to citizens as well as long-term sustainability (e.g., congestion, emissions, equity, and/or land use impacts). As TNCs are emerging as a rapidly growing urban mode of transport and represent the leading edge of mobility-as-a-service, TNC data availability has emerged as a critical data gap, and perhaps the most urgent. Addressing this gap will benefit Smart City analyses, and also provide the base data to extend urban travel behavioral models.
  • As a corollary to the previous take-away, the airport has emerged as the “front door” to Smart Cities, being the primary transportation hub welcoming visitors or connecting its citizens both to national and international air travel. As the rate of air travel growth far outstrips that of VMT (by about 3 to 1), the airport is emerging as a primary indicator of mobility behavior shifts, a sort of “canary in the coal mine.”  Collection and monitoring of airport access data provide critical insights into the rate of alternative mobility technology adoption. Similar changes are then anticipated to follow in other parts of the metropolitan area.
  • Lastly, and likely of greatest importance, a gap in practices that spans both the data infrastructure and travel modeling is the issue of appropriate metrics for smart city analysis. Mobility itself is typically not a smart city goal, but rather an enabler of improved quality of life. Metrics that quantify the ability of a city’s various transportation networks to connect citizens with the goods, services, and employment to increase the city’s productivity and improve citizens’ lives are needed for Smart City performance assessment.

While new technologies are enabling new data collection, modeling, and planning considerations, the research and science that can inform the future of cities need to keep pace. Data/model integration, visualizations, and analytics will continue to emerge, and a goal of this initiative is to further enable data-driven decision making through exchange of best practices via cross-city smart analysis such as this. Overall, this curation activity is intended to enable efficient access to the knowledge generated from Smart City peer cities, share knowledge and lessons learned, and benchmark their progress. It will also aid in continuing to identify gaps in knowledge and practices, which in turn will expose opportunities for the DOE EEMS initiative to contribute to Smart City objectives while gaining insight and valuable data from Smart City programs.


Castiglione, J., T. Chang, D. Cooper, J. Hobson, W. Logan, E. Young, et al. 2017. TNCs Today: A Profile of San Francisco Transportation Network Company Activity. Draft Report. San Francisco County Transportation Authority (SFCTA), June 2017.

Add to RSS Feed Add to Technorati Favorites Stumble It! Digg It!

Gerry Reid

Gerry Reid

“Technology leader with 20+ years experience in Agile IT Development, consulting, operations, delivery, project management in CRM, robotics, automation, cloud in financial services, telecoms & consulting sector", ReidAnalyticsSmart CitiesForeword The U.S. Department of Energy’s Energy Efficient Mobility Systems (EEMS) Program envisions an affordable, efficient, safe, and accessible transportation future in which mobility is decoupled from energy consumption. The EEMS Program conducts early-stage research and development at the vehicle, traveler, and system levels, creating new knowledge, tools, insights, and...CRM consulting and technology for Ireland and Europe, in the Public and private sector