3. Washington, DC Household Survey Key Findings

The purpose of this section is to address the first two questions on the second objective, namely:

• What is the share of travel that is made for various purposes other than commuting on congested freeways in the Washington, DC area? and
• What share of these travelers may have some flexibility to shift their time of travel to off-peak periods?

Methodology Used to "Mine" the MWCOG 1994 Survey

The MWCOG 1994 household travel survey is a statistically valid survey for the Washington region, with a sample of 4,863 households. One drawback is that the survey is more than twelve years old. (MWCOG is in the process of conducting a new household travel survey for the region. However, data will not be available until mid-2009, well outside the timeframe for this project.) The Study Team obtained the documentation and an electronic version of the MWCOG 1994 survey in ASCII comma-delimited format from MWCOG. The structure and definitions are similar to National Household Travel Survey (NHTS). Some "data cleaning" was required.

The following data sorts, combinations and analyses were performed on the survey data, as defined below:

1. Separated trip records by peak and non-peak and created trip chains for work and non-work travel. Work trips were defined as all trips with an ultimate origin or destination of work as long as interim stops are 30 minutes or less (to overcome the problem of a work trip with a stop to pick up coffee being coded or interpreted as a shopping trip). Work trips include home to work trips (HW), work to home trips (WH), work to work trips (WW), work to other trips (WO), and other to work trips (OW). Non-work trips include home to home trips (HH), home to other (HO), other to home (OH), and other to other. The morning peak period is defined as trips beginning from 6 am to 9 am, while the evening peak period is defined as trips beginning from 3 pm to 6 pm.[5]
2. The MWCOG survey did not include questions on distance, but did include travel time records. Trip origin and destination addresses were not available, however, information on the city or county of origin was provided. (The study scope and time frame precluded a detailed analysis by Transportation Analysis Zone or TAZ). Travel times greater than twenty minutes are used as surrogates for trips that are likely to use freeways. These are referred to as longer trips; the majority of the analysis focuses on these longer trips.
3. Identified mode of travel (personal vehicle versus all other; all other meaning rail and bus transit, bike, walk, and other) to focus on longer trips using a personal vehicle as the “target universe”
4. Evaluated longer AM Peak trips with stops on the way to work, including travel time from home and purpose (e.g., stopping for daycare or coffee) and longer PM Peak trips on the way home from work.
5. Compared AM Peak and PM Peak trip purposes and patterns for longer work and non-work trips using a personal vehicle, narrowing the focus further to those trips with 0 or 1 stops.
6. Analyzed longer non-work personal vehicle 0-stop or 1-stop trips to identify those discretionary trips that appear most likely to be able to shift time. The study defines discretionary trips based on trip purposes. The discretionary category includes family and personal business, social and recreational activities, shopping, and restaurant trip purpose destinations as discretionary.
7. Developed graphic representations of the data to summarize essential findings.

MWCOG Survey Findings

Note: All Findings are based on trip chains, not individual unlinked trips, so results are likely to vary from other studies. We begin with broad overviews of travel purposes and patterns, then evaluate subsets of the data to address the research questions as fully as possible.

• Who travels in peak and off-peak periods? How many trips are for work versus non-work? How many of these trips are longer trips (greater than 20 minutes)?

Figure 3-1: Trips by Time of Day and Overall Trip Purpose

Figure 3-1 demonstrates that there are very significant volumes of travel throughout the day and during peak periods, including large numbers of non-work trips. The three columns to the left summarize total trips during the day. Approximately 61 percent of the 2.37 million total AM Peak Trips are work trips, while slightly less than half of the PM Peak Trips are work trips. [6] Trips that take more than 20 minutes travel time (longer trips) are depicted in the three right-hand columns, and show somewhat different patterns from the total trips. For example, in the AM Peak, the 1.4 million longer trips comprise 60 percent of the 2.4 million total trips, while in the PM Peak the 1.5 million longer trips comprise 50 percent of the 2.9 million total trips.

Figures 3.2 and 3.3 illustrate the major differences between the AM Peak and PM Peak in terms of proportions of longer trips, and in proportions of work and non-work trips for these longer trips. The AM Peak and PM Peak total volumes of longer trips are very close to the same, at 1.42 and 1.46 million trips, but the proportions of work and non-work trips are quite different. In the AM Peak, over 1 million (72 percent of the 1.42 million longer trips, or 43 percent of all AM Peak trips) are work trips, while less than 400,000 (28 percent of the 1.42 million, or 17 percent of all AM Peak trips) are non-work trips. In the PM Peak, over 900,000 (63 percent of the 1.46 million longer trips, or 31 percent of all trips) are work trips, with almost 540,000 (37 percent of the longer trips, 18 percent of all PM Peak trips) non-work trips.

Figure 3-2: Key AM Peak Trip Characteristics

Figure 3-3: Key PM Peak Trip Characteristics

The remainder of the analysis will focus on these longer trips in the AM and PM Peak Periods, steadily narrowing the focus to identify the appropriate "universe" and our "target population" of longer peak period discretionary trips. The first step is discerning mode: private vehicle use versus other modes.

Among the universe of longer trips in the AM and PM peaks, the personal vehicle is the mode of choice for work trips and non-work trips, as shown in Figures 3.4 and 3.5.

Figure 3.4 demonstrates that personal vehicles account for 1.0 million trips or 71 percent of all longer trips in the AM Peak. Because the other 29 percent of the trips include walking, biking, and transit, as well as "other", we will consider only the personal vehicles as our "universe" for traffic for the remainder of the analysis. Once we eliminate the "other mode" trips, work trips represent 80 percent of the AM Peak longer trips.

Figure 3.5 shows a very different pattern for the PM Peak trips, compared with the AM Peak Trips. Personal vehicles account for 1.13 million of the longer PM Peak period trips, or 77 percent of the total. Of these 1.13 million trips, only 64 percent are work trips, with 36 percent non-work trips. It is also significant that the volumes of non-work personal vehicle trips in the PM Peak are more than double the volumes in the AM Peak.

Figure 3-4: AM Peak Longer Trip Distribution by Type

Figure 3-5: PM Peak Longer Trip Distribution by Type

Morning and Afternoon Peak Longer Discretionary Trips

One of the challenges for the study is to estimate the non-work trips that are using freeways during peak periods. Without a complete origin-destination matrix of trips, and without distance being part of the survey, we must make some broad assumptions about travel patterns in order to address this question. As noted in the introduction to the survey, since we do not have direct distance information, we are using time as a surrogate and assuming that many of the longer trips—more than 20 minutes each—are using freeways. For this "drilling-down" into the data, we ask the following questions:

• What are the key trip purposes for the longer non-work trips?
• Of these longer, non-work trips, which categories are commonly deemed to be discretionary and therefore capable of moving the travel to other time periods, other days, or other routes?

Many trip destination purposes are included in the survey, including Home, Work and Related, School, Child Care, Pick Up / Drop Off, Shopping, Family / Personal, Restaurant, Changing Mode, and Other. Activity Modeling analyzes work and non-work trips into three main categories: Maintenance, Subsistence, and Discretionary, as well as some trips that are generally not separately categorized in the analysis bridging trips and activities (marked as N/A – Home is self-defined)[7]. Work trips are considered in the Subsistence category. The key to the Trip Purposes delineated in the MWCOG survey for the following analysis and figures is as follows:

Abbreviation	Explanation	Category	Included as Discretionary?
Home	Non-Work-Home Trip	N/A	No
PckUpDrp	Pick up or Drop off	N/A	No
ChngMode	Change Mode	N/A	No
Other		N/A	No
School	School	Subsistence	No
ChldCare	Child Care	Maintenance	No
Shop	Shop (for goods or services)	Maintenance	Yes
Restrant	Restaurant	Discretionary	Yes
Fam/Pers	Family Personal	Discretionary	Yes
SocRec	Social/Recreational	Discretionary	Yes

Although shopping is identified as a maintenance activity, we believe that in general it may be considered as discretionary, particularly in terms of choice of time of day or day of week for most types of shopping, in particular for the longer trips included for this analysis. We analyze AM Peak and PM Peak travel in turn. In order to err on the side of caution in our estimates of discretionary trips, we focus our attention only on non-work trips with 0 or 1 stops, to reduce or eliminate the likelihood of identifying a discretionary trip with an interim non-discretionary destination purpose. Figures 3.6 and 3.7 identify the trip purposes for AM Peak and PM Peak non-work trips, longer than 20 minutes, taken in personal vehicles. In each figure, the four pieces of the pie that are extracted represent discretionary trips, as defined for this study.

Figure 3.6 illustrates the non-work trip purpose destinations for these longer trips in the AM Peak period, using a personal vehicle. After eliminating home, school, child care, etc. we have over 77,000 longer trips that can be identified as discretionary. This represents 38 percent of the pertinent non-work trips, and a full 7.7 percent of the total personal vehicle longer trips (work and non-work).

Figure 3-6: Non-Work AM Peak Trip Purposes

Figure 3.7 illustrates that, as expected, there are substantial differences in trip purpose distributions between the AM Peak and the PM Peak, as well as differences in volumes, as previously noted in relation to Figure 3.5. PM Peak discretionary trips total about 118,500, about 53 percent more than in the AM Peak. Though these discretionary trips represent only 29 percent of the PM Peak non-work trips, they account for a full 10.5 percent of all the personal vehicle longer PM Peak trips. This is due to the earlier finding, that non-work trips represent a much greater portion and volume of travel in the PM Peak compared with the AM Peak.

Figure 3-7: Non-Work PM Peak Trip Purposes

What Does This All Mean?

Longer discretionary trips form a substantial portion of the AM Peak and PM Peak trips. However it is not feasible to discourage or remove all discretionary trips from freeways, and even if it were, removing all such trips would not necessarily be sufficient to ensure free-flow conditions on all segments of the freeways.

Figure 3.8 illustrates how the discretionary trips, "teased out" of the non-work trips, form a notable portion (7.7 percent) of the AM Peak "universe" of personal vehicle trips. These trips have the greatest flexibility to shift time or route away from crowded freeway facilities; this shift can be encouraged through pricing and other incentives. Shifting this base of trips away from peak congested facilities would help congestion in many corridors; as noted in the traffic analysis section above, the reduction in traffic needed to restore "free-flow" conditions varies by facility, but in general will require a range of 10 percent to 14 percent to attain free-flow.

We have established that reducing traffic by 10 to 14 percent (in the PM Peak, generalizing in this case to the AM Peak) will result in a 75 to 80 percent reduction in delay, which would benefit millions of travelers every day. Achieving a 10 to 14 percent reduction in the AM Peak cannot be accomplished solely by diverting discretionary travelers. In addition to strongly encouraging AM Peak discretionary travelers to shift travel to other times or routes, it will likely be necessary to encourage other travelers to shift as well. Strategies to reduce work trips in personal vehicles, such as telecommuting or transit/ carpool/ vanpool incentives and facilitation, have also proven effective in the region, as shown by the 16 percent "other modes" for AM Peak work trips in Figure 3.4. More can be done; strategies that decrease travel times for carpools, vanpools and buses increase the appeal and market share of these modes, as discussed in Section 4. Thus, shifting AM Peak discretionary trips off our congested freeways, to other times or other routes, represents a promising strategy that must be used in concert with other strategies.

Figure 3-8: Summary AM Peak Discretionary Trip Breakdown

Figure 3-9: Summary PM Peak Discretionary Trip Breakdown

Figure 3.9 summarizes the somewhat more promising potential for reducing discretionary trips in the PM Peak. Discretionary personal vehicle trips making one or no stops represent a full 10.5% percent of longer PM Peak personal vehicle trips- within the target range of the 10 percent to 14 percent reduction required to achieve a 75 to 80 percent reduction in delay. Since PM Peak trip volumes are substantially greater than AM Peak trip volumes, and since this is largely due to the higher volumes of non-work trips, strategies that encourage discretionary trips to seek other times or other routes show great promise for reducing congestion in the PM Peak. Strategies that reduce commute trips by using pricing as a tool to promote carpools, vanpools, transit and telecommuting for work trips provide another significant “pool” of trips for reducing congestion, as discussed in Section 4.

---