View the November 20 seminar recording
Jennifer Symoun
Good afternoon or good morning to those of you to the West. Welcome to the Talking Freight Seminar Series. My name is Jennifer Symoun and I will moderate today's seminar. Today's topic is International Urban Freight (I-NUF) Conference Presentations, Part 1 – Curbside Delivery Challenges and Opportunities
Before I go any further, I do want to let those of you who are calling into the teleconference for the audio know that you need to mute your computer speakers or else you will be hearing your audio over the computer as well.
Today's seminar will last 90 minutes, with 60 minutes allocated for the speakers, and the final 30 minutes for audience Question and Answer. If during the presentations you think of a question, you can type it into the chat area. Please make sure you send your question to "Everyone" and indicate which presenter your question is for. Presenters will be unable to answer your questions during their presentations, but I will start off the question and answer session with the questions typed into the chat box. We will also take questions over the phone if time allows and I will provide instructions on how to do so once we get to that point.
The PowerPoint presentations used during the seminar are available for download from the file download box in the lower right corner of your screen. The presentations will also be available online within the next few weeks, along with a recording and a transcript. I will send a link to the recording in the next day or so and will also notify all attendees once all materials are posted online.
Talking Freight seminars are eligible for 1.5 certification maintenance credits for AICP members. In order to obtain credit for today's seminar, you must have logged in with your first and last name or if you are attending with a group of people you must type your first and last name into the chat box.
PDH certificates are also available for Talking Freight seminars. To receive 1.5 PDH credits, you will need to fill out a form. Please see the link in the chat box. Certificates will be emailed one week after the seminar. A seminar agenda has been included in the file download box for those who need to submit an agenda to their licensing agency.
Finally, I encourage everyone to please also download the evaluation form from the file share box and submit this form to me after you have filled it out.
Today we'll have three presentations, given by:
Our first presentation will be given by Giacomo Dalla Chiara, a Research Associate at the Urban Freight Lab, University of Washington in Seattle. He holds a PhD in Engineering Systems from the Singapore University of Technology and Design (Singapore) and a MSc in Statistics from ETH Zurich (Switzerland). In his work he develops models and simulations to study and develop new sustainable urban logistics practices.
Giacomo Dalla Chiara
Thank you, Jennifer. Good morning, everybody. The objective of this study that I am presenting today is to answer the question, "Do commercial vehicles cruise for parking?" Therefore, I will introduce a simple method that says whether we can start talking about cruising for parking for commercial vehicles. This is in collaboration with Professor Anne Goodchild, the founding director of the Urban Freight Lab at the University of Washington. I will start with a brief introduction and I will present the methodology we developed to assess parking cruising. Then we apply the methodology to GPS data, and conclude with some results.
So, let's start from a definition of what we mean by parking cruising. When demand for parking approaches parking supply available in urban areas, we start having what has been called a mobile queue of vehicles circling around the block and looking for parking; that is what we call parking cruising. There are two types of cost related to parking cruising. One is an internal cost: how much of my time I spend searching for parking. There have been studies that tried to estimate what this average search time is. There is also an external cost. Several studies found out that, on average, we have about 34% of the road traffic cruising for parking, and this increases the congestion on the road. Also, if I park for an additional hour in a parking lot, how many other vehicles do I cause to cruise for parking because they could not park? This is been evaluated by Shoup (2006), and by Inci, Ommeren, & Kobus (2017). Other research has focused on which factors affect parking cruising. One of these factors that was found is parking costs, parking prices, and the difference between on-street and off-street parking prices. Other studies have looked at parking information systems and traffic information systems and so on.
So, why are we interested in this topic? Why is it important to evaluate parking and cruising behaviors? What is the value of time? And, also, what is the patience of a driver who will spend time searching for parking? What is our on-average park-and-go time and what is our willingness to pay to find parking faster? All this boils down to having better vehicle parking policies. And all of the studies that have been done focused on passenger vehicles. It is very important to start talking about parking pricing, parking enforcement, minimum parking requirements for off-street parking, and time limits.
So, what about commercial vehicles? So far, most of the research that has been done was focused on passenger vehicles. When we talk about freight parking demands, what we know is that it has been increasing over the years. Diane, in the next presentation, can tell us more about that, giving some numbers. But, in general, because we know that there has been an increase in e-commerce sales as well as reduction in the volume of goods moved, then we expect parking demand to increase. What about parking supply? What about the infrastructure we provide to commercial vehicles to park and perform load and unload in urban areas? Here I have a pie chart of the curb allocation in downtown Seattle. In Seattle around 11% of the curb has been allocated to commercial vehicles (CVLZ). But in general, we do have a sense that parking supply is usually not enough for freight vehicles in urban areas. We expect commercial vehicles to have some troubles in searching for parking.
So, regarding the research gaps: as I mentioned, most of the literature that focuses parking cruising is on passenger vehicles. There has been literature that delves deeper into commercial vehicle parking. A common assumption when we do urban modeling is that commercial vehicles do not cruise for parking. Sometimes this is an assumption because including parking cruising in our models is complicated. Other studies describe that commercial vehicles have a tendency of parking in unauthorized spaces, such as double parking. They argued that, because there is illegal parking for commercial vehicles, we don't expect them to spend time searching for parking. But this is a topic that we believe we should spend more time digging deeper into the data and try to actually quantify this problem. Also, the fact that we have a lot of unauthorized parking does not mean that all commercial vehicles do not search for parking, or that they always park in unauthorized spaces. So, these are the research question we put forward. First of all, is there empirical evidence of parking cruising for commercial vehicles? And therefore, can we estimate what is the "internal cost", so the average time they spend cruising for parking? And then, lastly, what factors affect the parking cruising?
So, let's go to the methodology. Very simply, we have what we call a trip time. A commercial vehicle departs from a given shop or retail area and then arrives in another area in the city, and the total time from finishing unloading in one place and stopping the vehicle and start unloading in another place we call "trip time." We decompose it into two different parts: travel time and what we call "deviation". The travel time is the time the vehicle moves from A to B. We call "deviation" whatever is the remaining time. So, the parking cruising is included in the deviation, because once the vehicle arrives in a certain area, it has to look for a parking lot. The method we propose is very simple. First of all, we observe real trip times; we will describe afterward our sample of GPS data. We obtain reliable travel time estimations; those are the times a vehicle takes going from A to B, considering traffic conditions but not considering parking. And then therefore, the deviations are the difference between real trip time and the estimated travel time. Once we have the deviation, we want to check what the deviations actually are, and whether parking infrastructure affects the deviation.
What is the best estimate of travel time? In this research we used Google Maps' Distance Matrix API. It has some assumptions, it is not exact, but I think it is the best we can do at the moment. It does assume the fastest route, the output is the travel time estimation, and it considers historical traffic time conditions. Therefore, we can query the Google Distance Matrix API for specific day of the week, time of the year, and time of the day.
Then, I will describe the data sources that we used. First, we have trip data and the other source is the parking infrastructure data. We used around 2,000 truck trips from a parcel delivery company that delivers in Seattle. These trips were performed by 11 drivers over 28 weekdays. You can see in the picture on the map just a portion of a trip. For each trip we have trip start and the origin, the destination, and the GPS location. Then we want to correlate the destination of each trip with the availability of parking infrastructure at the destinations and other characteristics of the destination. Therefore, for every destination location we create a buffer with a radius of about 80 meters. And then we query the map and ask things like how much curb is allocated to commercial vehicles zones, how much curb is allocated to street parking, or how many private loading bays do we have in the buffer area? We also have data about parking occupancies. How many vehicles, for instance, have paid for parking? This data is available for Seattle.
So, we applied the methodology that I described earlier to the data described. We start with some descriptive statistics. We computed trip deviation, the difference between the real trip time we obtained from GPS data and the estimated travel time that we obtained from Google Maps. Then we obtained the graph you see on the left that gives you the trip time deviation distribution. What we see is that the most frequent deviation is zero. Which is actually a good thing because it means that we do estimate correctly the trip time, travel time most of the times is equal to the actual real trip time. We do observe some negative values which means that the truck's actual trip time was faster than the Google maps estimated travel time. But we also have the 70% you see on the right-end of the table, are either zero or greater than zero. On the right-hand side of the slide you see the cumulative distribution in which you can clearly see the median. So, we have a median of around 3.4 minutes, we use the median as a robust estimation of the average. We do have extreme values: I do not believe that there exist cruising times of 30, 40, 60 minutes. Those are problems that come with working with real data. But the best estimate we get is this 3.4 minutes of median. We then clustered all the trips ends destinations into groups, and we computed the mean cluster trip deviation. Then, we can look at the map. And in this case, we can see on the map of downtown Seattle, as we go toward the top right, we have darker dots; trip time deviations are increasing compared to the south part of the city. We can start thinking about what happened in that street compared to the street in the southside of Seattle.
Then we perform a quick check of what affect this trip time deviation. To do this we use regression analysis. We regress the logarithm of the trip time on the logarithm of the travel time while we control for other variables, including the curb allocate to CVLZs. In this case, when we estimate our betas from our regression analysis. For instance, the beta for the CVLZs corresponds to the amount of curb allocated to commercial loading zones in the buffer for a given trip destination. If we estimate the beta, and it is negative, that can be interpreted as the more curb allocated to commercial vehicles, then we have a shorter trip time controlling for travel time, and therefore, we have a shorter trip time deviation. We use different input variables in the regression analysis. We have three main trip variables: travel time; dwell time (so when the driver finished the trip, how long is he/she parked for); and time of day. We've used tour variables, because a stop is within a tour that a driver performed in a day. We have what time do we start the tour, how many stops are in the tour. We can quantify whether different drivers have different parking cruising behaviors. And parking variables: how much curb is allocated for different types of parking, and number of bus routes. We try different regression models; we have tried three main types of models. One is OLS (Ordinary Least Square), and we also tried mix effect models, because we have observations for several drivers and at the same time, we have destinations buffers that overlap.
Here are some regression results. We do have, of course, that travel time is significant and positive. Therefore, longer travel time increases trip time. This is significant in every model. When we look at the tour variables, what is interesting is that the most significant variable is the stop number in the tour, and it has a negative beta. That indicates that, as we advance throughout the tour, we see a reduction in deviations. We interpret this as a hurried behavior, in which we are less willing to search for parking toward the end of a tour. When we look at different parking variables, we have that the length of the commercial vehicle loading zone has a negative effect of the trip time, controlling for estimated travel time. As well as paid parking. We interpret this as more curb we allocate to commercial vehicles and paid parking, then we can reduce the trip deviation and therefore the time spent searching for parking. Other types of infrastructure provided positive effects for an increase in deviation.
Therefore, to conclude, we started with the question, "Do commercial vehicles cruise for parking?" What we found, given our small sample data, is that we have a nonzero trip time deviation with respect to estimated travel time with a trip time deviation median of 3.4 minutes and around 70% of the trip time deviations in our sample are greater or equal than zero. We also found the trip time deviations are statistically significantly affected by parking infrastructure provided at the destination.
We need to delve deeper into the data, and this data is not perfect. It comes only from a single parcel delivery company. We're looking to improve the study with increasing the amounts of data used, trying different supply chains, and recomputing for different data sources, using the same method, what we call trip time deviations, and see whether different supply chains have different parking cruising behavior. I would like to thank everybody for listening and also the FHWA for inviting me at this seminar.
Jennifer Symoun
Thank you, Giacomo. Our next presentation will be given by Diana Ramirez-Rios, a Graduate Research Assistant at Rensselaer Polytechnic Institute in The Center of Excellence for Sustainable Urban Freight Systems and The Center for Infrastructure, Transportation, and the Environment. Her research interests are in Urban Freight Transportation and Disaster Response Logistics. Her recent work in freight demand modeling involves the empirical estimates of freight and service trips in US cities, which provides a unique opportunity to quantify the impact of commercial vehicle traffic in urban areas. She is a fellow of the ENO Future Leaders in Transportation, MIT CEE Rising Stars, and was part of the team of finalists for the 2017 Franz Edelman Award who developed the Off-hour Delivery Program in New York City. In 2015 she was designated as a Colciencias Scholar by the Governor of the Atlantico Department in Colombia.
Diana Ramirez-Rios
Thank you, Jennifer. Hello, this is Diana Ramirez-Rios. I am here to present our research: On-street Parking Requirements for Freight and Service Activity. This is a work in collaboration with Professor Jose Holguin-Veras and Professor Lokesh Kalahasthi at the Center for Infrastructure, Transportation, and the Environment, and the VREF Center of Excellence for Sustainable Urban Freight Systems. During my presentation I will talk about the background, the freight and service trips, and the procedure we developed to estimate parking requirements in the cities, and results for selected cities in the United States. At the end I will brief you on some concluding remarks.
So, we know that cities, particularly the urban core, face numerous challenges with daily truck traffic. We see that this is most prevalent in the last leg of delivery where we observe road obstruction and we see all these kinds of issues like double parking and a truck in the intersection unloading and loading activities, because there is lack of space. Our previous research has estimated this issue and has illustrated huge parking fines that drivers have to pay, particularly in Manhattan. Some key observations about this issue is that there is a lack of sufficient freight parking and this leads to increases in double parking, increases in traffic due to cruising to find parking, and as a result congestion increases. And on the other hand, we have increases in the cost of goods due to the extra time it takes to find parking and the parking fines that have to be paid. And on top of this we have to account for the needs of service activity. When we talk about commercial parking, we not only include freight but also service vehicles, and in some cases also limos. And although service visits are less frequent, they have longer durations, so they have major implications in parking. How can we address this issue? Well, we have developed an estimate where we are able to quantify on-street parking requirements. And we would like to take further steps to allocate space where it is needed.
So, with respect to freight and service trips I would like first to give you a background on the key concepts that we are talking about in our study. When we talk about freight, we talk about freight generation, which is the freight production, the tons delivered and received by the establishment. And we know that freight generation is a manifestation of the production and consumption processes. So, we expect that freight generation will increase with increasing economic inputs. Then we talk about freight trip generation, which on the other hand is majorly responsible for logistical decisions. The freight trip generation are the vehicle trips that are produced and received by the establishment. And a FTG does not necessarily increase with the economic input, because shippers can decide on increasing or decreasing the shipment sizes and deliver more or less trips. And also, we have the service trip generation which are also vehicle trips produced and received by the establishment, but due to service visits. And in our research presented here we will be focusing on the FTG and STG models.
So FTG and STA, which is the service trip attracted, were developed based on establishment surveys that we developed as part of the NCHRP and NCFRP projects where our team has collected for more than 10 years data on deliveries received, shipments sent out, and service visits received. This model estimates and predicts the deliveries and shipments and the service trips made using employment as a variable. And we have developed approximately 1,000 models and these models are also based on industry sector, particularly the 2 digit and 3 digit NAICS code. And in addition, our team has developed a generation model for Internet deliveries to households. So, regarding our freight trip generation models we have a very consistent result comparing to some control points, some external data that we know from previous research, and they have shown consistent results with our models. And the advantages also on having the establishment level models and since we have them by industry sector, this has shown to be a better predictor than for instance the square footage of the establishment. Also, since we have employment-only models, we are able to use data from publicly available data, like the county business patterns data, and we are able to estimate models routinely. We also have the Commodity Flow Survey for freight production estimates. And if collecting data is desired, we can have smaller samples that work well with the NAICS because it creates homogeneous groups. And the models since they are at an establishment level, we are able to aggregate them. We can estimate if you want to know for example the buildings, how much freight and service they generate, and we can even go ahead and estimate for larger geographic regions. Our models also, as an effort of our research, we have developed a freight service activity generation software which includes all of our models and is able to process County business patterns data for different years and generate an estimate for all the different cities in the United States. And we are able to estimate as well at a fiscal level and then we can estimate, county, city, and the state levels. And we have also produced an estimate for 2 digit and 3 digit NAICS. If you're interested in using our software, it is publicly available. Just copy the link and go to our software and you will have a page like this where it will ask your user name and password. There are instructions so you can send us an email and we can provide you the information you need to login.
So, to show you some results. We analyzed of 8 different American cities of different sizes. These were selected based on population density. So, we see New York City as the largest one and then we can go down to Kansas City, Kansas. And as shown in our results, we have estimated freight trip generation, service trip attraction, and also internet deliveries to households. And we can see this as an indicator per capita at a per mile basis, and these are all daily deliveries. Typically, our results show that service trips are approximately 10% of the freight trips, and if we include internet deliveries to households, then the total deliveries and shipments per capita ranges from 0.23 to 0.29. The numbers are big. And our estimates can also be observed to show where in a geographical region is the bulk of these trips and where is the challenge. In the case of New York City, we can see the bulk of trips are in lower Manhattan, and this is shown from our zip code based estimate. We can also go further if we have available data at an establishment level and we can analyze blocks. In this case we had information available. These results, we didn't have information on Grand Central Terminal, but we were able to show some other LTGs, so we can see where the freight and service trips are generated in these blocks.
So further, we want to understand the parking requirements at the cities. What we did was we selected the most congested zip codes at each city, and we estimated at this zip code the daily freight and service trips, and then we were able to obtain parking requirements using the number of trips and approximate duration of the delivery. And then the number of parking slots were generated that were required at the zip code level. For example, in New York City we have the largest one that contains the Empire State Building and other large buildings. Here, only this zip code has close to 24,000 people and over 8,000 establishments with nearly 175,000 employees. From these we are able to generate that only this zip code generates 35,000 FTGs and nearly 4,000 deliveries to households and 2,400 visits. Another example of this is part of the downtown of Seattle, the most congested zip code has nearly 3,000 establishments and 87,500 employees. And for this variable also to estimate the deliveries and service trips. In general, for every zip code we were able to generate these estimates for the most congested zip code at each of the cities. And we are able to understand how many trips and services are generated at a per mile basis. And to understand a little bit more this number we produced this graph. We are able to see that if all the deliveries and shipments in a day were placed next to each other, we were able to calculate how much space would be between them. So, in the case of like smaller, smaller cities that have a huge space like Kansas City, Columbus, and we have a lot of space between the deliveries. But in the case of New York City it is only four feet per delivery. This might sound like an exaggeration, but actually this is not a new problem. It is a reality in the 1940s. It shows that precisely, there is a lot going on, and this block is pretty close to downtown Manhattan.
So, for each of these zip codes we are able to calculate traffic at peak hours. If there were 25% of the total daily traffic in the peak hour, we were able to estimate the space available. We assumed that the freight trip lasts around 20 minutes and the service trips last one hour and a half. Actually, the hour and a half was a value that we estimated with our own research, as well, and the service duration is approximately 1.5 hours. It can be longer, or it can be shorter, but that is the average. And using this estimate we are able to calculate all the space required at each one of these zip codes. And we are able to estimate how much space per 100 feet is required. So, we see the bigger cities like New York having a worse lack of space compared to the other cities. And if the peak hour was reduced, using for example some management initiatives like off-hour deliveries, we can reduce this number even lower and we can also better solve the problem of insufficient parking at these cities.
So, some concluding remarks on our results. This research estimated parking requirements for freight and service trips. It takes advantage of the FTG-STA models. And the results indicate that depending on the level of density of the cities, there is a wide range of parking requirements at each case. And peak hour parking needs at the most congested zip codes show how insufficient road space for parking can be. And by reducing peak hour traffic one can reduce this value considerably. And as bad as it is, the situation is bound to get worse with the effects of the on-demand economy. We want to note that these result does not contemplate on-demand, like food deliveries, that are increasing more in the recent years; these results do not contemplate this aspect. With that I would like to thank you for inviting me and thank you for listening.
Jennifer Symoun
Thank you, Diana. Our final presentation will be given by Scott Strelecki, who has been a Senior Regional Planner at the Southern California Association of Governments (SCAG) since late March 2018. Prior to joining SCAG, Mr. Strelecki worked at the San Diego Association of Governments (SANDAG), beginning his transportation planning career in May 2008. His core focus over the previous seven years has been on regional goods movement and transportation finance. Recent project work has included in-depth review and analysis of commodity flows throughout southern California, and analysis and research on last-mile freight delivery, e-commerce, and parcel and package delivery supply chains. Prior to working in the public sector, Mr. Strelecki held positions at law and investment firms from 2000 to 2006. Mr. Strelecki received a Master of City Planning from San Diego State University in 2008.
Scott Strelecki
Thank you, very much and good morning, everyone from out here in Los Angeles. Today I will make a presentation on the last mile freight study which is a work effort we have been undertaking over the last year and a half or so, and the ultimate objective of today's presentation will be to focus on how this study is looking at developing an assessment of freight curbside activities to have a framework to move forward in the regions here in Southern California.
Before going through this study, I would like to give a high-level overview on some of the trends that are driving the activity for last-mile deliveries. As all of us are aware and we are pretty much users of smart phones and there has been a strong empowerment that has occurred between the consumers having more transparency, better ability to compare prices and make quick decisions, having more access to information. And currently this has resulted in putting some pressure and leading to opportunities for innovation in the private sector for looking at how companies ship products and how carriers can provide services to meet those customer expectations. E-commerce, as was mentioned in the earlier presentation, continues to be a big driver in impact in both from a consumer perspective as well as in supply-chain. If we compare the e-commerce numbers to in-store retail sales, we see an increase from below 7% in 2008 to approaching 15% as of last year. The growth rate has tended to be 2-3 times that of retail sales over the last decade, or more, with much stronger performance even during recessionary pressure cycles. For the future, we are looking at the growth going from around 15% to approaching 1/4 of all retail sales. And when we break this down by who are the key drivers of that, Amazon always stands out at the top being the predominant driver, and then there are some other companies that make up an additional 15%. But the key takeaway is that if we look at the top 15 companies, currently they are driving about 70% of e-commerce as of last year. There's a lot of fragmentation as well, but there are a lot of specific companies that are kind of the driving force of this phenomenon. Ultimately what we would like to highlight is the distinction that there are so many different ways now of how goods can be delivered, whether it goes to a residence or business. So, this is a visualization of the supply chain aspect. We understand that based on trade or domestic production, there are traditional components looking at warehousing in distribution, but at the same time those have shifted toward fulfillment centers and urban delivery. And then just the number of how these goods can get to consumers at the last mile is changing dramatically, whether it is literally going to locations or there is consolidation of where the goods can be picked up. At the same time, we also want to highlight transportation network companies and ridesharing that is growing in the U.S. We recognize that the users continue to grow at a high rate. What is very interesting is looking at some of the global surveys that are out there. We look at the younger generations and demographic groups that are growing up with these types of smart phones and apps and technologies, and we are definitely seeing a much higher interest in using these types of services moving forward.
So ultimately, the goals of the study were to improve the regional understanding of delivery conditions, understand challenges and needs from a variety of users, quantify delivery issues and conditions, look at conflicting demands and attempt to develop strategies appropriate for different users, and identify potential concepts. The study was based in Los Angeles, but at the same time the goal was to tailor it in a way for strategies and opportunities that could lend themselves to other parts of Southern California. One of the most important parts was having a stakeholder-driven process. The elements were broken down into the following: obviously, stakeholder input; citywide data analysis; definition of typologies; and ultimately getting toward the case study location where data could be collected. Solutions were focused on literature review throughout the United States, case study recommendations, pilot project concepts, and a toolbox of strategies. For stakeholder input, a project advisory committee was formed, as well as interviews were undertaken with both delivery companies and receivers. Stakeholders played a major role in interpreting the data and the approach to help form the structure of a lot of components of the study. Ultimately, there was a lot of collaboration on project pilot concepts. So, we're highlighting some of the key stakeholders as you can see, major companies like Sysco, U.P.S., other parcel delivery companies, Target, but we also had a strong role in working with some academic institutions and contributing their insight, as well as working directly with the Los Angeles Department of Transportation and the city planning department.
So, for citywide data analysis, the key goal was to understand from the city the overall city standpoint of the numerous blocks there were, how to create parameters to screen that down into a manageable level of case study locations. Ultimately, the focus was on different screening parameters and street typologies in order to accomplish that. So, when the analysis was performed for street typologies, we can see the information, there is a predominant majority of residential with the remaining being commercial, industrial, and out/in service routes. This study, the focus lends itself predominantly to those minority areas as looking at the last-mile deliveries, as we wanted to have a focus area and if we tried to get into the residential typology that was going to create a substantial amount of time and effort. So, when looking at the screening parameters we looked at things such as truck volume, collisions, parking citations, number of deliveries, and those sort of categories. And if you break it down by typology, we can see that industrial major had a higher propensity toward truck volume and collisions whereas the commercial typology had higher parking citations and delivery demand. So, for the data collection effort, we screened parameters as you can see on the left map, as well as the tabular information below. A data collection plan was created, and sample collection analysis was tested to ensure that the approach would be most efficient. And ultimately, data was collected at 35 blocks and an analysis tool is available for any stakeholders who are interested on that data set.
So, just to walk through a little bit of that, each block type was organized by the specific curb uses at that block. On the left you can see red zones, driveways, parking (whether metered or not), crosswalks, yellow zones, all the different types that could be at each location. Ultimately these attributes were all incorporated into a database so that we could have the ability to do different types of analysis to help inform recommendations. As part of that there was a decision point to think about whether using video versus technician to collect the data. There were limitations and initial thoughts on whether either could be used independently, or whether both should be pursued. So, the study ended up using both video and technician. With some of the challenges with video related to site visibility and the perspective that one camera would provide, there was a need to consider multiple camera locations for each block which led to a limit of blocks that could use camera technology based on the budget for the study. Where the initial technician fieldwork showed that there was not an overwhelming aspect of collecting the data and the activities collected could be verified with video documentation. And there was not much additional time to tabulate the data, and then there was cleaning and reconciliation required as part of that. This is just an example that shows the equipment that was deployed on the blocks. You can see it doesn't stand out too much, so it wasn't too challenging to get the equipment up there to collect the information. Ultimately, we had 8,500 unique records collected across all the case study areas. And, with the ability of having each block location, we were able to match different types of characteristics, whether it was from the screening parameters and/or some of the challenges or issues that were identified as part of that. I should mention about a little over half was video based. And the primary benefit of video is higher-level type analytics versus the technician data being more beneficial for doing a little deeper dive into granular information.
So, the next couple of slides gives us a perspective of that, looking at the high level initially for the data records that were collected. This table is showing us the number of actions per day based on the total data that was collected and if we look at a couple examples, we can see the white zone actually had the highest number of actions per day with predominately parking and passenger activity, which is what we would expect with that with a small portion of deliveries. Whereas, if we look at the yellow curb type spaces having the second highest action per day, with a high amount of park usage there with a fairly high amount of delivery; we would expect high delivery usage. So, that is kind of giving us overarching type characteristics of the curb types. The other important attribute to consider was duration. So, every time a vehicle parked whether it was there briefly or stayed there a long time, the activities that were being done, whether it was passenger unloading, drop off/pickup, or an actual delivery being made, all those characteristics had a duration of a certain time associated with them. So, when we look at a high level park information, no surprise that the time is much longer than passenger drop-off and pickup and/or delivery, especially for the parking areas. If we exclude parking, the average duration drops significantly in other curb type uses. Passenger activity was very brief, which is expected, and deliveries averaged about 30 minutes per delivery.
At the more granular level using the technician data, we were able to look more deeply at the specific attributes. This table is showing us the distinction between delivery vehicles versus personal vehicle versus the trucks, your larger construction or van type trucks versus utility and other truck types. I think the highlight here that is important to note, of the delivery vehicles being the majority, that was reflected by mostly package parcel (FedEx, U.P.S.). There is no surprise, especially for the last mile. But having a high number of personal vehicles making deliveries was notable, as well as the high usage of the red zone for the majority of these vehicle types using that curb space. We also were able to get TNC information, and as part of that we noted that of all passenger loading, 10% were TNC. If we excluded bus activity, that amount doubled to about 20%. Similarly of note was the use of a red curb type for the substantial majority of those TNC drop-offs and pickups. This is just an overarching snapshot of what is not surprising to us, that the majority of deliveries are occurring during business hours. But what was also an issue was the fact that many of these routes began during morning peak and end in the evening peak time. So, there is a large issue with congestion related to actual component of these deliveries and many of the companies are tasked with trying to plan these routes and deliveries around that.
So, for the study recommendations we came up with three tiers including block level recommendations, pilot project concepts, and recommendations more for citywide level and toolbox of strategies. The tier 1 case study recommendations. Each case study area incorporated the different characteristics of the area as well as all the data collections that were attained. As an example, for this case study, area number one, we look at some of the recommendations on the right, we can kind of see how the data was incorporated where the usage of the white zone lent itself higher from a duration perspective, especially to commercial deliveries and then also for some passenger vehicle activity, which is one of the predominant uses of the white zone. So, that was potentially offered as a suggestion to be considered at potentially looking at converting to a yellow zone. And similarly, the red zone location in this area where it looked like there was enough red zone to maintain a portion versus converting a portion to yellow delivery area based on activity. That was the benefit data, looking at the local levels trying to identify where some potential low hanging fruit opportunities were.
And tier 2, looking at the toolbox of strategies. A I mentioned earlier, an extensive literature review was performed considering all of the different strategies and opportunities and innovations that are being performed across the country. These findings were categorized into three general areas looking at the curb area, looking at delivery companies and receivers, as well as applications and implementation. So ultimately some visualizations were created. Some depictions of those are to the right, looking at loading zone enhancements or coding the curb type opportunities, but ultimately a matrix was created trying to identify based on issues how solutions could potentially be provided with each area. So, an example of whether we are looking at an adequate curb loading, we could see that there could be many different options or considerations that a city or area could consider; versus if we were focused on per se an inadequate building loading, there were fewer very clear options that looked like there could be a connection to providing solutions to some of those issues. To a degree, some of this was subjective. We did try to incorporate the data to compare what we found. So, this is an opportunity to continue to improve on these type of relationships.
And lastly, looking at the pilot project concepts, there was good engagement with a lot of key stakeholders for both public agency for the city of Los Angeles as well as many different companies. I believe we interviewed close to 30 different receiving-shipping companies, as well as carriers of the goods. Ultimately, we came up with this is just some of the top concepts that we came up with through those discussions as well as doing some of the research on what other parts of the country and what they've been doing. The ones highlighted in blue are actions that we are having current discussions with key stakeholders and looking at potentially moving forward with in the near term as opportunities presenting themselves. I should mention that off-peak delivery is not highlighted but is part of that process.
Lessons learned; ultimately, we categorized into three areas. Prioritization is important. This study started as a last-mile delivery study. However, it very quickly became apparent that looking at curbside activities there are many competing uses and activities that occur. So, really focusing on what the priorities are depending on the use, and you can't be siloed, we have to consider the curb and its entirety and the benefit for all users. Optimize, really focusing on the data, the information that we can get to best help create transparency, better visibility. Information, looking at issues as well as strategies. And collaboration, obviously, considering to work with all the key stakeholders and moving forward on that front.
So, for next steps. We have been fortunate to get proceeds to do a curb space management study which we will look to build from our last freight study. The key focal points will be to expand the study from the city of Los Angeles to other counties within this region, as well as to revisit and enhance the data collection framework. Our efforts included video and technician, so we really want to look at more innovative opportunities where we can potentially scale this to collect much more robust data. And obviously, to continue to support the projects and other implementation strategies. So, with that, I believe that is the end of the presentation. I will turn it back over to Jennifer.
Jennifer Symoun
Thank you, Scott. We will now move on to the Q&A session with questions that of been posted online. Please continue typing in questions if you have them. Once we get through everything typed in the chat box, I will open up the phone line for questions.
I'm going to start from the top with some questions for Giacomo. The first question for you is, do you get good enough location information to say whether the vehicle is parked in a see CVLZ?
Giacomo Dalla Chiara
Thank you, Jennifer. So, what we have is latitude longitude coordinates regarding their true destination. We didn't try to match those coordinates with the respective parking choice. We didn't include in our model the actual parking choice, where they choose to park, because we thought that it was not precise enough. But we are working with some new data, not presented here, that will include parking choice, and that could tell us whether when they park illegally, when they double park, if they don't cruise, or whether they cruise.
Jennifer Symoun
Thank you. Another question for you, Giacomo. Did you consider the location of truck routes as well? What impact does it have on the deviation?
Giacomo Dalla Chiara
No, but that's a good idea. We can include that in the next iteration of the model. In the map that I showed in which we have different roads and we have different colors representing the different mean cluster trip deviation, we do actually have the pattern that for the routes that are preferred by trucks, those are more likely to have larger deviations, but we didn't include that in the regression model.
Jennifer Symoun
Another question for you. Since Google provides historical average travel time, how do you know that the differences you are observing due to cruising and not random variations due to changing local traffic conditions from day to day?
Giacomo Dalla Chiara
That's a good question. We do not know. Google Maps gives us a good estimates of travel time. Of course, the best estimate would be to be in a car and follow the truck at same time and same day and record the travel time. But that's impossible. So, that is what we have at the moment. That is also why we use regression analysis, because this is the way we check whether trip time deviations are somehow affected by parking infrastructure at destination, and therefore whether they could be interpreted as parking cruising. That is also why we talk about deviations, because we cannot say for sure whether those are truly parking cruising. Another problem that we encounter is what we call re-routing. For instance, it could be that a vehicle goes in a certain area but then it cannot find parking, therefore the driver decides to go to the next area to do the delivery and then comes back later on. In this case, the extra travel time is not caused by parking infrastructure at the destination but is caused by parking infrastructure in another place.
Jennifer Symoun
Thank you. One more question for you. Where do you get driver behavior data?
Giacomo Dalla Chiara
We have a driver ID. So, we know which driver performed which trip. And therefore, if we include that in our regression analysis (and we did that by testing for random intercepts) we can question whether different drivers actually have different deviations, different cruising time. And they did, in fact. And that is what I meant by driver behavior. However, we don't have variables like driver experience or other characteristics of the driver.
Jennifer Symoun
Thank you. Diana, some questions for you. How are you calculating the dimensions of packages deliveries to determine the number of trucks?
Diana Ramirez-Rios
Our data collected is not, I want to clarify, it's not on the pieces received or delivered. It is more on the deliveries made in terms of like numbers of trucks sent or numbers of truck received. So, that is the basis of estimating the freight trucks and service trips. We don't need to consider pieces. Although in the cases of deliveries of these types of one vehicle coming in and that estimating your trip, does not consider other conversion factors that probably add higher delivery densities. We should also consider if one truck is making multiple stops. So, that is a consideration we don't have here.
Jennifer Symoun
Another question for you. Concerning the on-street parking analysis presentation. Out of curiosity, why was Kansas City, Kansas and not neighboring Kansas City, Missouri, chosen for analysis? The latter is largest.
Diana Ramirez-Rios
Yes, Kansas City, Missouri is larger and probably has a lot of interesting results as well. But the selection of Kansas City, Kansas was because of the size of the city. We wanted a small size and that city fit the criteria when we developed our selection.
Jennifer Symoun
Another one for you. Do these models take into account off-street loading infrastructure?
Diana Ramirez-Rios
No, and actually that is a good point. Our analysis does not consider off-street. And obviously if we also account for off-street capacities, the numbers will reduce. Our numbers were mainly on-street.
Jennifer Symoun
Thank you. We have another question in here. I will put this one out to all of the presenters. Where do you see innovation helping in improving the whole experience of delivery, for example, drones? Scott, would you like to start off with that one?
Scott Strelecki
Sure. First off, I think there's a ton of innovation going on out there, from automated delivery for groceries to drones to robots, and I think a lot of these have different challenges as well as opportunities. So, the way we feel is that really having an understanding of what is going on in the environment, being prepared, and trying to test pilot project opportunities, that will be some things we are looking at. Any of these new innovations could fit within that, especially in Southern California. Looking intently at emission reductions and things like that. We are definitely trying to talk to all the people who are out there who are interested in trying to test these things and get some kind of quantifiable, measurable information to help inform policy and the process and the things that are going on.
Jennifer Symoun
Thank you. Giacomo or Diana, do you want to add anything to that?
Giacomo Dalla Chiara
This is Giacomo. From my perspective, new modes that do not have problems with parking, or less problems, like cargo bikes, are interesting to test. Another thing we're testing here at the Urban Freight Lab is providing parking information. And the question is if you provide parking information to drivers or to schedulers, can you reduce parking cruising or other problems with parking?
Jennifer Symoun
Diana, any thoughts from you?
Diana Ramirez-Rios
I agree with Giacomo and that the cargo bikes are a good innovation that has been provided with better opportunities to reduce the impact of pollution in deliveries. Also, there are other types of innovations that are not necessarily high-technology, but for example like consolidating deliveries in urban areas, having all these delivery boxes that are available are also good for reducing the number of trucks going into households. So, that is a good opportunity to reduce the impact of deliveries.
Jennifer Symoun
Thank you. Another question I will put out to all presenters. You may have seen results DC published last week in the collaboration with Curb Flow. Do you have any thoughts on the feasibility and impact of real-time curb reservations on the freight system?
Giacomo Dalla Chiara
This is Giacomo. I believe I can address that. We are not testing parking reservation. In a separate project we are testing parking information systems. Reserving the curb is an extra innovation which could be interesting to test.
Jennifer Symoun
Anything from you, Diana or Scott?
Scott Strelecki
I will add that I think we have heard a little bit of that and it is something we will definitely pay attention to and look into as well, and see if there is any opportunity to test something like that out here in Southern California.
Jennifer Symoun
Diana, anything?
Diana Ramirez-Rios
Not on my side.
Jennifer Symoun
All right. Are any of the speakers familiar with FedEx's Roxo, and what you think about urban robotic delivery?
Giacomo Dalla Chiara
This is Giacomo. I think those types of pilots are interesting. And I believe what we expect is similar to the cargo bikes. These vehicles will often not travel on the road nor they will park on the curb, they will often go on sidewalks. But they might bring different types of problems like congestion on the sidewalks, increasing the interactions between passengers and cargo bikes, passengers and robots.
Jennifer Symoun
Anyone else?
Scott Strelecki
Yeah, I think you have a lot of different companies looking into this. Doordash I know is sampling some of this for restaurant deliveries; Amazon, we all know is obviously looking at every opportunity that there is aside from drones. So, yeah, I think it just adds to complexity. You know, today there are a lot of companies that the way they operate, you have to just kind of deal with the complexity. But when we think of autonomous vehicles, different communication with infrastructure, these robots for the most part are probably right now going to be doing a lot of shorter type deliveries, not 20 miles or anything like that. It just gets really interesting once you start thinking of the current environment which is already complicated and all these new things that could potentially be looked at in the future.
Diana Ramirez-Rios
This is Diana. I also agree with the complexity that robots apply in the curbs and on the sidewalk. There are a lot of companies bringing this up and more of on the delivery for restaurants; it is a good opportunity for food delivery. But we have to account for that complexity of the conflict between other users of the space.
Jennifer Symoun
Thank you. Another question. How will freight lockers or consolidation sites affect your results? Again, anyone who wants to jump in and answer that.
Diana Ramirez-Rios
This is Diana. I think it's a good opportunity for freight lockers, a good opportunity for reducing the amount of deliveries made by trucks. If you can reduce the number of truck trips just like consolidating, it requires a huge effort of collaboration between stakeholders and also collaboration from the consumers, if they are willing to walk to a locker and pick up their deliveries. But it's a good opportunity to reduce.
Jennifer Symoun
Anyone else?
Giacomo Dalla Chiara
This is Giacomo. I agree with Diana. That consolidation and reduction of truck trips is very important. We observed the lockers is that they also reduce significantly the dwell time, the time a vehicle is parked in a parking place. If we reduce the dwell time, we can increase the parking turnover and therefore have more curb available for more vehicles.
Jennifer Symoun
Thank you. Scott, anything?
Scott Strelecki
I don't think I need to add much, I think I agree. The opportunity is there. I think with the lockers there's a little bit of differentiation between if it's an Amazon locker, for instance, other competitors aren't necessarily going to want to be in that environment, versus if it's that an office area or residential area where maybe it can be multitiered with all different companies being able to utilize it. But, in general, I think the consolidation opportunities can reduce potential trips or looking for parking, those sort of things.
Jennifer Symoun
Thank you. One more question for everyone. Have any of you considered or have done research with Uber freights data?
Scott Strelecki
This is Scott. I can maybe address that. I think in general, when we think about data sharing, all operators, owners of the data have the best primary data. I think in Uber's case for last-mile delivery, the predominant amount of data they have may not relate to last mile. It may be more truck load related. So, if it's maybe a business-to-business last-mile you could argue that, but not necessarily your package delivery type information. I think some of the major package delivery companies and Uber freight, you can put them in there too, they are maybe a little bit interested in working with you with some data, but they are very kind of cautious as far as like what you can get. Obviously, if anything, it will be much more aggregate versus detailed. I think there is maybe some willingness there, but I think it is a challenge. That is what we have experienced lately.
Jennifer Symoun
Thank you. Giacomo or Diana, anything from you?
Giacomo Dalla Chiara
No, I did not work with Uber freight data.
Diana Ramirez-Rios
Me, neither.
Jennifer Symoun
Okay, another question. In your experience with working with the local municipalities and elected officials, what messages helped you most when building partnerships among these topics? Anyone can start off on that.
Diana Ramirez-Rios
This is Diana. I think it's a very good question. One of the main things that we have experienced in our research is that the numbers are the ones that talk. That's important to convey to the public sector and that is what also provides the attention of them to work on projects and collaborate with us on improving the freight system in urban areas. That is the main message.
Jennifer Symoun
Anyone else?
Giacomo Dalla Chiara
This is Giacomo. At the Urban Freight Lab at the University of Washington, we have collaborations with private companies, and we bring them together with public institutions to address these problems. And we found that the curb is a very good common ground in which we have clear problems that the city can address. So, yes, bringing together both private carriers and the building managers and the city is very important. And talking together you get very interesting discussions.
Jennifer Symoun
Thank you. And Scott, anything you want to add?
Scott Strelecki
Based on the fact that we have kind of recently finished the draft version of our study, we are beginning to have some of these conversations. And I think that with the data, when we can look at specific blocks and we can highlight, for instance, you've got a bunch of TNC stopping in the red zones here to do their activity or you have a portion of deliveries that are occurring in red zones in these different areas throughout the city, I think highlighting those types of things really gets city Council districts engaged and understanding, kind of like having something to react to. And that gets the dialogue going and eventually if there's an interest you can move toward a really positive attitude toward looking at testing things. I think that is where we are at the forefront of beginning a lot of those conversations in the city of Los Angeles.
Jennifer Symoun
Thank you. We have a few minutes left. It looks like we might have some people still typing in questions. In the meantime, I will also get a chance if anyone wants to ask a question over the phone. You can press star five and I will open up your phone line for you. Again, if you want to ask a question over the phone press star five and I will open up your line.
We do have another question that was typed in. What is the thinking of truck parking at Highway Patrol inspection facilities? I will let anybody respond to that.
Giacomo Dalla Chiara
This is Giacomo. In the study we focused on downtown, dense areas. So, we don't have any data or results on highways.
Jennifer Symoun
Alright, Scott or Diana?
Scott Strelecki
This is Scott. Kind of the same answer; we didn't really consider that or focus too much on that. I'm not sure if there's a distinction between like a border crossing area or whether trucks have to be checked and there is capacity. But for the most part I think if it's a dedicated parking facility, specifically for trucks and/or if they have their different private lots and areas where equipment is stored. I'm not too sure about that, sorry.
Jennifer Symoun
Diana, anything from you?
Diana Ramirez-Rios
In our case we also have developed studies on mostly urban areas. We have also studied corridors to show the estimate on trucks traveling, like for example the I-87 corridor and from different major metropolitan areas. But we haven't considered the trucking parking on the highways or how this would work. Maybe that.
Jennifer Symoun
Thank you. I don't see any other questions typed in and I don't see anything over the phone. I will go ahead and start closing out but if you think of a question please feel free to continue typing them in and I will go back to the questions. I do want to thank all three of the presenters from today's seminar and as well thank everybody for attending. I will send out a link to the recording of today's webinar within the next day. The December Talking Freight seminar is not yet available for registration but once it is I will send notice through the Freight Planning LISTSERV. The Freight Planning LISTSERV is the primary means of sharing information about upcoming seminars. I also encourage you to join the LISTSERV if you have not already done so. In addition, if you are looking to obtain a certificate for 1.5 PDH credits please fill out the form at the link shown on the screen. A certificate will be emailed to you by the end of next week. AICP members can get their CM credits by logging into their account on the AICP website. With that, thank you everybody and enjoy the rest of your day and happy Thanksgiving to everybody.
