Evaluating an E-Scooter Pilot for Greensboro, NC

Evaluating an E-Scooter Pilot for Greensboro, NC September 2019

Table of Contents t ts

1. Introduction: Scooter Origins 2. Ridership Snapshot: Trip Highlights 3. Spatial Patterns: Scooters on the Map A. Trip Volume

Fig 1. Total Trip Volume Heat Map Fig 2. Most-Travelled Road Segments

B. Origins and Destinations

Fig 3. Destinations: User Drop-offs Fig 4. Origins: User Pick-ups

C. Trip Flows D. Rebalancing

Fig 5. Rebalance Drop-offs Fig 6. Rebalance Pick-ups 4. Scooter Safety: Eyes on the Road Fig 7. Time of Day Fig 8. Speed Limit 5. Temporal Patterns: Scooters by the Minute

Fig 9. Week-over-Week Trip Counts: Day vs. Night Fig 10. Month-over-Month Trips Fig 11. Average Trip Times Fig 12. Weekday vs. Weekend Trips

6. Conclusion 7. Methodology and Sources 8. Acknowledgements

c@1 STAE

Greensboro

greensboro-nc.gov

stae.co

1. Introduction: Scooter Origins

In November of 2018, the Greensboro City Council approved an ordinance and permit process to allow for the operation of shared e-scooter programs in the public right of way. Shared scooter operators received approval to deploy up to 200 scooters during a short-term pilot to test the long-term viability of this new form of micromobility. Bike and e-scooter company, Lime , was the only operator to participate, deploying vehicles throughout the pilot period: January 28th to August 1st 2019.

This report gives an initial look into the pilot by exploring data that Lime was required by permit to provide to the City via an MDS-compliant API ( Mobility Data Specification ).

In order to ingest and analyze this data, the City of Greensboro enlisted the support of Stae , a software platform that empowers cities to manage real-time, civic data. Stae served as the City’s data platform throughout the pilot, helping the City access, monitor, map, and analyze scooter data on a day-to-day basis, as well as to perform the analysis and findings shared here. Together, we began to explore trends and insights from this scooter data by asking simple questions such as: What were the most traveled routes? What time of day did people scoot? Was scooter usage higher on weekdays or weekends? Eight months of data can be overwhelming. We hope that through these simple maps and graphs, key stakeholders from City Hall and members of the public alike will be inspired to take an empirical approach to important discussions about the next steps for the scooter program. We created the following data visualizations using MDS and the Stae Mobility Data Manager platform.

2. Ridership Snapshot

Over eight months, how much did people use the scooters?

Scooters

Trips

Hours 14,668

200*

69,038

* Fleet size increased to 250 after July 1

What did the average trip look like?

Average number of daily trips: 371 trips

Median trip time 7 minutes

Average trip

time:

13 minutes 30

seconds

When were scooters in highest use?

Busiest month: 04.2019 13,340 trips

Least busy week: 03.04.19 to

Busiest week: 04.29.19 to 05.05.19 3,828 trips

Busiest day: 02.07.19 755 trips

03.10.19 (UNCG Spring Break) 795 trips

3. Spatial Patterns: Scooters on the Map

Where did Greensboro residents scoot to and from? These visualizations illustrate where Greensboro residents rode their scooters and emergent patterns of usage.

A. Trip Volume Fig. 1. Total Trip Volume Heat Map

This heat map depicts the routes of every scooter trip taken during the pilot period. Road segments in white represent areas of the street grid that were most frequently traveled. Segments in pink were the next most frequently traveled. The purple segments were less frequently traveled. For the purpose of protecting privacy, routes that were travelled fewer than three times have been excluded from this map.

Fig. 2. Most-Travelled Road Segments

The most traveled roads were concentrated downtown and near the UNCG campus. Below we see a map of the 500 most frequently traveled road segments, all shown in purple. These segments represent routes on which over 755 journeys were traveled during the course of the pilot.

B. Origins and Destinations

The most common origin and destinations were also downtown and on college campuses.

Fig 3. Destinations: User Drop-offs

The below heat map uses hex binning to show areas where trips most frequently ended. Areas with fewer than three trip origins are excluded from this map for the purpose of protecting privacy.

Layer Legend

Rider Dropoff Color by Point Count

1.00 to 1.00 1.00 to 1.00 1.00 to 2.00 2.00 to 4.00 4.00 to 8.00

8.00 to 19.00 19.00 to 71.75 71.75 to 3738.00

© Mapbox © OpenStreetMap

Fig 4. Origins: User Pick-ups

The below heat map uses hex binning to show areas where trips most frequently started. Areas with fewer than three trip destinations are excluded from this map for the purpose of protecting privacy.

Layer Legend

Rider Pickup Color by Point Count

1.00 to 1.00 1.00 to 2.00 2.00 to 5.00

5.00 to 14.00 14.00 to 61.33 61.33 to 4246.00

© Mapbox © OpenStreetMap

C. Trip Flows

“Flows” describe the volume of trips traveled between two zones. Looking at flows shows us in more detail where people are traveling to and from. The below map shows these trip volumes between different areas of Greensboro. Flows shown in white depict more frequent scooter travel between two areas, flows shown in pink are the next most frequently traveled, flows shown in light purple are the next most frequently traveled, and flows shown in dark purple are less frequently traveled.

© Mapbox © OpenStreetMap

D. Reblancing

Rebalancing is when shared vehicles are moved by the operator from an area of relatively low demand to an area of relatively high demand. Rebalancing can also be used to promote policy goals, such as equitable access to vehicles across neighborhoods. You can see Lime rebalancing the scooters below.

Fig 5. Rebalance Drop-offs

The below map shows hex bins with counts of the number of times Lime dropped of a scooter in this location.

Layer Legend

Rebalance Dropoff Color by Point Count

1.00 to 2.00 2.00 to 3.00 3.00 to 5.00

5.00 to 10.00 10.00 to 48.33 48.33 to 2992.00

© Mapbox © OpenStreetMap

Fig 6. Rebalance Pick-ups

The below map shows hex bins with counts of the number of times Lime picked up a scooter in this location to move it elsewhere.

Layer Legend

Rebalance Pickup Color by Point Count

1.00 to 3.00 3.00 to 5.00 5.00 to 9.00

9.00 to 19.00 19.00 to 59.50 59.50 to 1868.00

© Mapbox © OpenStreetMap

4. Scooter Safety: Eyes on the Road

Safety is a critically important part of riding new shared e-scooters—especially in the context of initiatives like Greensboro’s Vision Zero . We can look at several factors to see how Greensboro is doing on scooter safety, including time of day and speed limit.

Fig. 7. Time of Day

As seen below, the busiest times of day are during commuting times and in the evening after work. During the pilot, 17% of rides took place at night (defined here as between the hours of 10 p.m. and 6 a.m.), when visibility and other factors could potentially cause a safety concern. 00h 01h 02h 03h 04h 05h 06h 07h 08h 09h 10h 11h 12h 13h 14h 15h 16h 17h 18h 19h 20h 21h 22h 23h Mo Tu We Th Fr

Sa Su

≥ 0

≥ 100

≥ 200

≥ 300

≥ 400

≥ 500

≥ 600

≥ 700

≥ 800

Fig. 8. Speed Limit

By ordinance, scooters are limited to streets with speed limits under 35 mph. In the below map, we can see where scooters went on streets that were legally too fast for them. The yellow lines show streets where the speed limit exceeds 35 mph, while road segments highlighted in blue show segments where scooter trips took place. Areas of overlap between these two might warrant scrutiny for greater enforcement or for improved infrastructure, such as a dedicated scooter lane.

It is not possible to decipher from the data whether these trips took place on the street or on the sidewalks of these streets, which is also prohibited in accordance with Section 16-222 of Chapter 16 of the Greensboro Code of Ordinances, “Motor Vehicles and Traffic”.

5. Temporal Patterns: Scooters by the Minute

How did scooter use change over the course of the pilot? In this section we’ll dive deeper into scooter patterns based on time of year and time of day.

Fig. 9. Week-over-Week Trip Volumes: Day vs. Night

The below graph shows that rides during the day make up a majority of trips. The y-axis shows total number of trips and the x-axis shows the start date of a given week of the pilot program. The darker purple represents trips that took place between the hours of 6 a.m. and 10 p.m., while the lighter purple shows “night trips”—those that took place between the hours of 10 p.m. and 6 a.m. You can hover over each bar in the graph to get more precise counts. If trips were evenly distributed throughout the day, we would expect one third (33%) of all trips to take place during the 8 hours of the 24-hour day that we have defined as “night trips” (between the hours of 10 p.m. and 6 a.m.). As we see above, night trips make up less than a third of total trips for nearly all weeks shown. These trips make up 17% of the total trips throughout the course of the pilot.

4,000

3,000

2,000

1,000

Total Number of Trips

0

1/28/2019

2/4/2019

2/12/18/2019 1/2019

2/25/2019

3/4/2019

3/13/18/2019 1/2019

3/25/2019

4/1/2019

4/8/2019

4/15/2019

4/22/2019

4/29/2019

5/6/2019

5/13/2019

5/20/2019

5/27/2019

6/3/2019

6/10/2019

6/17/2019

6/24/2019

7/1/2019

7/8/2019

7/15/2019

7/22/2019

7/29/2019

night

day

Fig. 10. Month-over-Month Trips

We can see the number of scooter trips increasing as the weather gets warmer. Trips peaked in April & May towards the end of the school year and remain high during the summer months.

12,000

10,000

8,000

6,000

Number of Trips 4,000

2,000

0

January Febuary

March

April

May

June

July

Fig 11. Average Trip Times

Although the number of trips is lower in the summer months than in the spring, we can see that the trip time increases as the weather gets better, peaking in June at over 10 minutes per ride.

July June

May April March

Febuary January

0

2

4

6

8

10

Time (minutes)

Fig 12. Weekday vs. Weekend Trips

As spring turns to summer, we also see the average number of weekend rides per day surpassing the number of weekday rides per day.

600 200 300 400 500 100 0

Total Number of Trips

1

1

1

1

1

1

Feb

Mar

Apr

May

Jun

Jul

Weekend

Weekday

6. Conclusion

We hope the above exploration of Greensboro’s scooter data can be useful to decision makers, transportation stakeholders, and members of the general public alike. This preliminary analysis is by no means exhaustive, and we encourage anyone interested to check out the source data—an aggregated version of the trip data shared by Lime— on Stae to conduct their own analysis and continue informing this important community conversation.

7. Methodology and Sources

Data are from the Data Mobility Specification provided by Lime to the City of Greensboro and hosted on Stae. The dataset for the pilot period was downloaded from Stae to be used for analysis, first filtering for: 1) trips that took place between January 28, 2019 12:00 a.m. and January 2, 2019 12:00 a.m. and 2) scooter trips (excluding bike shares), and 3) trips longer than 100 meters. These data were further cleaned by checking for duplicates using the trip ID field, and dropping trips with fewer than three latitude or longitude coordinates (the primary reason for this threshold was to eliminate errors that showed trips straight across the City with very few coordinates, so those). Trips longer than 2.5 hours were also dropped. The data cleaning and analysis for trip date and time fields was done in R and Python. Dummy variables were made to indicate additional trip attributes such as whether a trip took place at night, on the weekend, and which day of the week. The time zone was adjusted to be on EST and additional attributes were calculated such as trip duration, and total and average trips. Data were reformatted to be grouped and summarized by day, month, and week, which were used to make the Britechart graphs. The spatial analysis was done following the SharedStreets methodology for Mobility Metrics that snaps trips to the street line, bins origins and destinations, draws ridership “flows” across the city, and calculates route frequency. We used the cleaned and re- formatted data from this process to create maps in Kepler.

8. Acknowledgements

Our thanks to the following partners in the City of Greensboro::

Chris Spencer , PE, Engineering Division Manager, Department of Transportation Chandler Hagen , Transportation Planner / Bicycle and Pedestrian Coordinator, Department of Transportation Jane Nickles , CIO, Information Technology Department Rodney Roberts , Deputy Chief Information Officer, Information Technology Department

Special thanks our Civic Data Design Fellow, Lydia Jessup for her work researching and creating the visualizations for this report.