shapeimage_3.png

Preliminary Patronage Forecast for a Napa/Solano Rail Transit Network

publictransit.us Special Report No. 10

Michael D. Setty • Leroy W. Demery, Jr. • December 8, 2005

Copyright 2003–2005, Publictransit.us

1.   Executive Summary

According to an increasing body of academic research, short-distance travel by bus is much less popular with the general public than by rail transit (which in turn is less popular than travel by personal motor vehicle, not an unexpected result). Though bus service is less popular with the traveling public than rail transit, improved bus service is still often needed in markets where ridership potential does not meet the minimum passenger traffic density thresholds needed to offset rail transit’s much higher capital costs.

In 2003, the Napa County Transportation Planning Agency (NCTPA) and Solano Transportation Authority (STA) completed the Napa/Solano Passenger/Freight Rail Study. This study resulted in what this report calls the “R.L. Banks Reference Rail System,” which was projected to carry about 6,200 and 6,500 weekday passengers in 2010 and 2020, respectively. Routes would operate between Vallejo and St. Helena, Vallejo and Fairfield/Suisun City, and Napa to Fairfield/Suisun City. Capital expenses were estimated for separate services, but a composite system was estimated to cost $216 million (2003 dollars), exclusive of right of way purchase. Operating expenses for all services evaluated totaled $24.8 million per year, with no allowance for economies of scale for a combined system.

Shortcomings of the “ R.L. Banks Reference Rail System” included a failure to consider impacts of traffic from outside the service area, an arbitrary, second unnecessary distinction between “commuter” and “tourist” service, failure to consider all-day service and a number of other major flaws in the study’s underlying assumptions and analysis.

This report concludes that a 62-mile, two route rail passenger system between the Vallejo Ferry Terminal, the Napa Valley, Fairfield, Suisun City and Vacaville could attract up to 25,600 daily riders by 2010 and 26,700 daily riders by 2020, based on Napa and Solano County travel projections previously used by the 2003 Napa/Solano Passenger/Freight Rail Study. In turn, the mathematical model on which the projections in this analysis are based was developed from BART and Caltrain patronage data collected for an East Bay BART extension study.

Key assumptions include: (1) commute period service frequency every 15-20 minutes on both routes, except every 30 minutes north of Napa; (2) all-day service every 15-30 minutes; (3) connecting ferries to San Francisco expanded to meet potential demand, plus extensive express bus “bridges” to BART along I-80 and I-680 across the Carquinez Strait; (4) “commuter rail” technical standards to minimize system capital expenses, but also providing sufficient track capacity so that a “transit” level of service can be operated; and (5) “state of the art” hybrid battery/diesel-electric locomotives or railcars with acceleration that matches BART railcars. No attempt at calculating operating expenses was made in this analysis.

“Traffic density” describes the number of passengers who, on average, travel over each kilometer (or mile) of system length. The average trip length per passenger for the Napa-Solano rail scenario is projected to be about 16.0 miles in both 2010 and 2020; on a typical weekday, passengers would travel about 4 2 0,000 and 4 35 ,000 miles, respectively. The system’s overall passenger “traffic density” is projected to be about 6,800 and 7,012 “daily passenger miles per route mile” (dpm/rm) in 2010 and 2020, respectively. Federally-funded research from the early 1980’s, and independent Japanese and European studies, indicate that as a general rule, the minimum “threshold” of feasibility for rail transit is a traffic density of 5,000 dpm/rm. BART carries about 40,000 dpm/rm, and Sacramento light rail, 7,700 dpm/rm. For the Sonoma/Marin "SMART" commuter rail project, the authors project 2,500-3,000 dpm/rm.

Research by Pushkarev, Zupan and Cumella (1980) also estimated that initial transit capital expenditures of $1,250 per daily passenger mile per route mile (19 77 dollars ; approximately $ 4 ,000 in 2005 dollars) were justified based on estimated travel and other benefits. Based on this yardstick, system capital expenditures of up to $1.7 billion could theoretically be justified. “Commuter rail” design standards could hold this estimated expense to $400-$500 million. A rail extension “Up Valley” north of Napa would not be justified using the “traffic density” threshold. However, many passengers would be tourists, who could be charged higher fares to offset the projected traffic density shortfall.

This analysis also suggests that a rail transit connection between BART in Contra Costa County and the Vallejo Ferry Terminal would be very productive, and could generate passenger traffic density in the range of 15,000–20,000 daily passenger miles per route mile.

1.  Consumer Preferences and Transportation

Rail transit is often proposed to obtain a wide range of expected benefits. As Todd Litman explains in the abstract summary of his recent study Rail Transit in America: A Comprehensive Evaluation of Benefits (2005) ( www.vtpi.org/railben.pdf ):

This study evaluates rail transit benefits based on a comprehensive analysis of transportation system performance in major U.S. cities. It finds that cities with large, well-established rail systems have significantly higher per capita transit ridership, lower average per capita vehicle ownership and annual mileage, less traffic congestion, lower traffic death rates, lower consumer expenditures on transportation, and higher transit service cost recovery than otherwise comparable cities with less or no rail transit service. This indicates that rail transit systems provide economic, social and environmental benefits, and these benefits tend to increase as a system expands and matures. This report discusses best practices for evaluating transit benefits. It examines criticisms of rail transit investments, finding that many are based on inaccurate analysis.

Recent academic research also indicates a generally strong public preference for rail transit relative to buses. This research strongly corroborates earlier analysis by Tennyson (1989) and Demery et al. (1994-2005). In May 1998, UC Davis professor Patricia L. Mokhtarian and her academic collaborators (Ory and Mokhtarian, 2004) surveyed 1,358 San Francisco Bay Area residents regarding attitudes towards short distance and long distance travel by personal vehicles, buses, rail transit, and walking, jogging, and bicycling. The most important findings are summarized in Table 1 below.

Table 1.  Short Distance Travel “Liking” *

Mode

Strongly Dislike

Dislike

Neutral

Like

Strongly Like

Personal Vehicle

2.5%

9.2%

30.2%

47.6%

10.5%

Bus

28.6%

34.8%

28.3%

7.6%

0.7%

Rail

11.9%

17.0%

39.8%

28.3%

3.1%

Walk/jog/bicycle

4.0%

4.9%

24.4%

48.8%

17.9%

A total of 58.1% of those surveyed “liked” or “strongly liked” short distance travel by personal vehicle, compared to only 8.3% by bus, 31.4% by rail, but 66.7% by walking, jogging, or bicycling. In contrast, only 11.7% of survey respondents “disliked” or “strongly disliked” personal vehicle travel, 63.4% disdained travel by bus, 28.9% weren’t enamored of rail transit, and only 8.9% failed to appreciate walking, jogging, and bicycling. The largest proportion of those surveyed (39.8%) were “neutral” towards travel by rail, while 24.4% to 30.2% were neutral to all other travel means.

The survey sample was drawn primarily from three San Francisco Bay Area neighborhoods generally familiar with personal vehicle, bus, and rail travel in the region: (1) “North San Francisco, ”(2) Concord, and (3) Pleasant Hill. All three subareas are served by the Bay Area Rapid Transit (BART) system. North San Francisco is also served by the extensive bus, trolley coach, cable car and streetcar service provided by the San Francisco Municipal Railway (Muni). In addition to BART, Concord and Pleasant Hill are served by much less extensive, far less frequent bus service provided by the Central Contra Costa Transit Authority (CCCTA).

These survey results provide a persuasive explanation for the empirical findings by Demery et al. (199 4 -2005) sho wing that rail transit attracts about 35% to 45% more passengers that buses relative to capacity provided. That is, substantially more people are “neutral”, “like” or “strongly like” rail compared to buses; as a result, transit users as a group tolerate more densely-packed loads on trains than they are willing to on buses.

The extensive empirical data collected by Demery et al (1994-2005) indicates that it is highly unusual for bus systems during peak periods to carry more than 1.0 to 1.2 passengers per foot of vehicle length (about 3.0–3.5 passengers per meter of vehicle length). This is about 40–50 passengers on a 40-foot bus, and 60–70 passengers on a 60-foot articulated vehicle.

In contrast, passenger loads of 1.2 to 1.5 per foot of vehicle length (4.0–5.0 passengers per meter of vehicle length ) is relatively common on rail systems. According to Demery and Higgins (2003), many new light rail transit systems experience relatively high passenger loads per vehicle compared to buses:

Post-1990 data for 17 corridors in 12 LRT systems (Table 2b) suggest that the 100 p/v benchmark falls at the median peak vehicle occupancy carried by recently-constructed LRT facilities. In terms of passengers per meter of vehicle length, the median, 4.0 p/m, is just below that for HRT and corresponds to 104 passengers per 85-foot (26-meter) vehicle. The highest peak vehicle occupancies among the 17 corridors are 4.9 p/m in Portland, 4.7 p/m in Sacramento and 4.6 p/m in Calgary. The Portland figure approaches that for Boston's Green Line LRT system: 5.1 p/m (1994).

A 1989 paper by Edson L. Tennyson, P.E., Impact on Transit Patronage of Cessation or Inauguration of Rail Service, concluded:

In most cities served by buses exclusively, transit riding has declined 75 percent over the past 40 years [as of 1989]. Exclusive busways have not made much difference absolutely, but they have helped relatively. In 11 areas with updated rail transit facilities, ridership has increased markedly, often by more than 100 percent.  In two of these areas, the transit systems are attracting more ridership than they did when gasoline and tires were rationed. It appears that rail transit makes a great difference in ridership attraction, with attendant benefits

Because transit use is a function of travel time, fare, frequency of service, population, and density, increased transit use can not be attributed to rail transit when these other factors are improved.  When these service conditions are equal, it is evident that rail transit is likely to attract from 34 percent to 43 percent more riders than will equivalent bus service. The data do not provide explanations for this phenomenon, but other studies and reports suggest that the clearly identifiable rail route; delineated stops that are often protected; more stable, safer, and more comfortable vehicles; freedom from fumes and excessive noise; and more generous vehicle dimensions may all be factors.

Given the extensive evidence that rail transit is considerably more attractive to would-be riders than bus service only, rail service in Solano and Napa Counties warrants additional evaluation, at least in corridors where potential transit patronage is sufficient to warrant the high capital costs of new rail transit.

Though bus service is less popular with the traveling public than rail transit, improved bus service is still often needed in markets where ridership potential does not meet the minimum passenger traffic density thresholds needed to offset rail transit’s much higher capital costs.

2.  The 2003 “R. L. Banks Reference Rail System”

In 2002 and 2003, the Napa County Transportation Planning Agency (NTCPA) and Solano Transportation Authority (STA) conducted an extensive, $250,000 study of rail passenger and freight issues in Napa and Solano Counties. This study, authored by consultant R.L. Banks & Associates and associated subcontractors, concluded that weekday peak hour only, hourly commute service (e.g., 4 schedules each morning and 4 schedules each evening) could be provided on three routes while not interfering with then-existent heavy freight traffic. These routes of what we will refer to as the “R.L. Banks Reference Rail System” were: (1) Vallejo Ferry Terminal–Fairfield/Suisun City; (2) Vallejo Ferry Terminal–Napa–St. Helena; and (3) Fairfield/Suisun City–Napa.

In addition to this quite limited commute hour service proposal, two midday round trips each from the Fairfield/Suisun City and Vallejo terminals could be provided to the Napa Valley, catering to tourists. Patronage was projected using three different methods, resulting in a “composite estimate of commute patronage summarized in Table 2 below.

Table 2.  R.L. Banks Reference Rail System: Commute & Visitor Patronage Estimates

Corridor

2010

 

2020

 

Weekday Commute AM & PM

Daily

Annual

Daily

Annual

Vallejo – Fairfield/Suisun City

1,732

436,600

1,934

487,200

Vallejo – Napa - St. Helena

1,984

500,000

2,324

585,400

Fairfield/Suisun City - Napa

1,586

399,420

1,754

441,800

TOTAL, Commute Patronage

5,472

1,433,664

5,842

1,472,700

Tourist-Oriented Services*

       

Fairfield/Suisun City - Napa

308

112,480

308

112,480

Vallejo – Napa - St. Helena

382

139,520

382

139,520

TOTAL, Tourist Patronage

690

252,000

 

252,000

GRAND TOTAL

 

1,685,664

 

1,724,700

*  Two highest results for each mode indicated in bold .

The R.L. Banks study estimated capital costs of $99.4 million to $138.6 million (2003 dollars) for “stand alone” rail routes, e.g., operating each of the examined routes as a separate entity. Capital costs of $216 million (2003 dollars) was estimated for a system encompassing all three routes. Annual operating expenses (2003 dollars) were projected to total $24,835,000 for the five services examined in the study. It is important to note that the R.L. Banks study did not evaluate operating expenses of a combined rail system, e.g., where overhead, administration, track and facilities maintenance and other relatively fixed expenses were shared, e.g., reducing overall expenses. Since these expenses would be incurred only once as part of the “economy of scale” inherent in rail technology, overall operating expenses would be several million dollars per year less than the expenses totaled in Table 3.

Table 3.   R.L. Banks Reference Rail System: Expenses, Revenues, Cost Effectiveness

Route

Annual Unlinked Trips

Operating Expense

Projected Revenues

Projected Operating Subsidy

Estimated Passenger Miles

COMMUTE SERVICE

         

Vallejo – Fairfield/Suisun City

454,046

$4,760,000

$1,147,000

$3,613,000

7,423,953

Vallejo – Napa - St. Helena

519,808

$6,931,000

$1,000,900

$5,922,000

4,881,000

Fairfield/Suisun City - Napa

459,810

$4,881,000

$983,000

$3,898,000

11,052,994

VISITOR SERVICE

         

Vallejo-Napa (Rutherford)

139,520

$4,017,000

$439,000

$3,578,000

4,632,064

Fairfield/Suisun-Napa

112,480

$4,246,000

$354,000

$3,892,000

4,454,208

GRAND TOTAL

1,685,664

$24,835,000

$3,923,900

$20,902,000

27,568,100

Source: Napa/Solano Passenger/Freight Rail Study, pp. ES-6.

Based on the R.L. Banks analysis, the “stand alone” route rail system would recover about 16% of its operating expense from fares–without consideration of the significant savings of a combined system as discussed earlier. Assuming a combined system costing about $17 million to $18 million annually to operate, farebox return could increase to 23% to 24%.

Ironically, the projected capital cost effectiveness of the R.L. Banks Reference Rail System is well within the criteria proposed by the seminal transit study Urban Rail in America: An Exploration of Criteria for Fixed-Guideway Transit (Pushkarev et al., 1982). In the Urban Rail analysis, the median estimated capital expense per average daily passenger mile (DPM) or U.S. rail projects then in development was $1,250 (1977 dollars), roughly $4,000 (2005 dollars).

The estimated $216 million (2003 dollars) capital cost for the R.L. Banks Reference Rail System–divided by 90,000 weekday passenger miles (27.6 million annual pass-mi. divided by 300 days per year) results in a capital cost effectiveness of $2,500 per DPM, approximately 60% of the updated Pushkarev el al measure. This relatively low cost is not surprising, given the proposed use of relatively inexpensive commuter rail technology and not including the costs of right-of-way purchased from the Union Pacific Railroad or Napa Valley Wine Train.

3.  Criticisms of the R.L. Banks Reference Rail System

In the view of the authors, the R.L. Banks rail proposal has many shortcomings, which are summarized in this section.

Failure to consider traffic from outside the study area is a serious oversight.

A high-quality commuter service, with good connections to San Francisco ferry service at Vallejo, may be expected to attract some traffic from southeastern Sonoma County and southern Lake County. Impacts (e.g., parking at stations) may be significant. Although not great by volume, traffic from southern Lake County might double the commuter patronage carried by an extension from St. Helena to Calistoga.

“Commuter rail” and “tourist rail” is a false dichotomy.

To report that some communities support "tourist rail over commuter rail," or vice versa, begs the question of what the difference is.

A rail service providing “substantial benefit” by “significantly reducing congestion” or “providing a faster and more convenient transportation alternative” -- one likely to attract public support for financing -- would obviously need to serve as many markets as possible. This includes commuters as well as visitors. As a practical matter, service operated for one market would also serve the other (absent some measure to physically bar visitors from commuter trains, which is totally absurd).

As a practical matter, it would not be possible to serve one market segment (commuters) to the exclusion of the other (visitors) simply because some passengers would find it convenient to use "both" service categories. Some visitors would use rail even if it were confined to peak-period, peak-direction service. As another example, a commuter needing to make a midday trip from work to home would obviously benefit from the availability of midday service.

The visitor-oriented or “tourist rail” service proposal, with its emphasis on a “gateway” station, shuttles and limousines, has evidently been drafted with little input from potential consumers. The proposal appears to reflect less of market research regarding the wishes of visitors themselves, and more the influence of the “pull-up-the-drawbridge” mentality prevalent in some Napa Valley communities. It is one thing to determine where visitors to the Napa Valley come from, where and how long they stay, and so forth. It is quite another thing to determine where and when they wish to go while visiting the Valley, and what type of service might attract them out of their cars.

The market segment that would be served by the “visitor oriented service” would resemble that already served by escorted tours and the Wine Train. In fact, major functional differences between the Wine Train and the proposed “visitor oriented service” are few: (1) the Wine Train does not provide transportation into and out of the the Napa Valley; (2) the "visitor oriented service" does not provide meals onboard trains.

Competition between the two is less of an issue than the potential, unrealized under the “visitor oriented service” proposal, to serve additional segments of the "Napa Valley Visitor" market. Some visitors may wish to travel along the Highway 29 corridor, stopping and visiting wineries and other attractions on a casual basis.  The Draft Final Report recognizes this in Chapter 6. However, it then recommends a "package" approach, including a one-way train ride, with “a shuttle bus touring local attractions” in the opposite direction.

If “stop and visit” is the “typical” travel pattern of visitors arriving by private auto, then the proposed rail “visitor” service will attract fewer passengers than would be possible otherwise. The rail line is well-located to serve most of the attractions in the Napa Valley, with a minimum of connecting shuttles. The R.L. Banks study confuses the issue by stating that “many visitor destinations are not within walking distance of proposed stations.” A large share of these, however, could be served by well-located additional stations.

“Casual” travelers would be more likely to use the service given all-day, two-direction service with an hourly interval between trains, and a handful of additional stations to better serve various attractions. Benefits to Napa Valley residents and businesses include reduction of visitor-generated auto traffic.

The Loire Valley in France provides an example of what can be accomplished with local rail services. It is true that the Loire region is larger (Orleans to Nantes is 200 miles), with a larger population, than the Napa Valley. However, local rail services, every hour or two during the day, permit visitors to travel along the valley as they wish, stopping at various attractions, particularly between Orleans and Tours, about 70 miles. An article about such travel in the Loire, by writer Zane Katsikis, was published by International Railway Traveler during the late 1990s.

FROM: Chapter  4: “Recommended Station Plan”

Section 4.5, “Station spacing and its impact on operational productivity.”

Increasing the number of stations may make the service convenient to more potential customers but also will lengthen train schedules and may discourage potential riders.

This “fundamental tradeoff” is true in principle, but its real-world impact may not be significant. Consumer choices result from consumer perceptions, attitudes, and so forth, not from the outputs of ridership forecasting models. Aside from trained observers, few people can perceive a two-minute change in the length of a 20-minute interval (i.e. the likely average duration of ride) without reference to a timepiece.  The effect of a single additional station stop would go unnoticed by most passengers.

The influences of "wait time" (or "out-of-vehicle time") on ridership is not independent of average travel distance.  As average travel distance grows, the influence of “wait time” diminishes because it accounts for a declining share of total travel time. The influence of schedule speed on ridership is also not independent of average travel distance. Given a 12-mile average travel distance, the difference between 30 mph and 40 mph accounts for 6 minutes of travel time, just barely significant. However, given a 36-mile average travel distance, the difference becomes 18 minutes, which is strongly significant.

If each additional station stop adds two minutes to end-to-end running time, 2-3 stops could be added (4-6 minutes additional end-to-end time) before some passengers -- those making longer trips -- would notice. Additional well-located stations, together with a reasonable all-day service level, would increase the overall usefulness of the service and thereby attract additional patronage.  As the Draft Final report notes:

 . . . a new service should be designed to meet the transportation needs of its constituents rather than to slavishly conform with standard practices.

It is neither necessary nor reasonable for trains to stop every mile in the manner of the old electric interurban railway. On the other hand, various passenger railways in Japan routinely add new stations to serve new development. Benefits of additional stations, e.g. increased ridership and revenue, offset disadvantages from marginal travel-time increases.

Adding a stop in outlying area where fewer riders are on board the trains affects a lesser number of passengers than adding a stop near inbound destinations at a point where more or most passengers are on board.

In the case of 1-2 added stations (= 2-4 minutes additional time), this is unlikely to have any significant effect on ridership. The effect becomes noticeable if running time extended by more than about 5 minutes. There should be a minimum threshold for travel-time savings and related issues, otherwise meaningless “results” may drive public policy. This “minimum” might be six minutes (i.e. 0.1 hour), as this is the "minimum increment of time" charged by many businesses and professions.

In addition, where bi-directional service is operated, lengthening running time could cause a need to acquire a need to acquire additional train sets and train crews to execute the service plan.

This type of thinking is typical of small- and medium-sized bus transit operators, and causes endemic problems with service quality. Because schedules cannot be kept owing to traffic congestion, connections between lines are often missed -- in particular, between bus lines of two different operators.  This certainly discourages ridership to some extent, but the effect would be much greater if a larger share of bus passengers had access to alternative transportation. The Draft Final Report makes the point that the estimated St. Helena --- Vallejo travel time, “about 55 minutes,” does not leave “sufficient time and cushion” to make a one-way trip, turn and depart within the planned 60-minute headway.  The report notes that more train sets and crews would be required to provide the 60-minute headway than if the one-way trip time were “a few minutes less.”

The above is an example of "false economy." The share of potential passengers likely to travel on “commuter” and “visitor” services who have access to alternative transportation will be much higher than those who travel by current transit services. The additional “trainsets and crews” would provide a needed “cushion” to insure service quality (that is, providing a “cushion” against service disruptions and other factors, such as larger-than-normal loads that might slow service owing to longer dwell times). Slightly longer running times would also permit addition of additional stations which might prove desirable as traffic grows. Additional equipment and staff would also facilitate operation of special service for special events (an important tool for mitigation of auto traffic attracted by special events in other areas).

. . . rail systems require longer acceleration times to stop at stations than do buses and require longer dwell times at stations.

“Longer dwell times” are not an inherent characteristic of rail systems. Nor are marginally slower acceleration and braking rates a significant disadvantage given the potential advantages of rail service from the passenger's perspective, including greater comfort, greater reliability and higher average speed.

FROM: Chapter 6: “Commuter and Visitor Ridership and Revenues”

. . . requiring riders to transfer may assure that all possible riders will use the train, yet the additional time and effort involved in transferring keeps some from wanting to use the transit system altogether.

Experience by transit operators with well-developed "timed transfer" arrangements demonstrates that additional patronage generated by greater connectivity (i.e. making transit a viable option between a larger number of origins and destinations) more than offsets whatever is lost by the need to transfer per se.

“Increas[ing] rail ridership yet decreas[ing] the overall market of transit riders” is a theoretical construct -- popular among certain rail opponents -- that takes no account of experience in other cities, and quickly becomes untenable given any reasonable attempt to account for changes in travel volume (passenger-miles).

. . . Vallejo Transit bus routes which parallel the rail line are currently experiencing strong farebox recovery . . .

Farebox recovery for Vallejo Transit route 90/91 (Fairvield/Vacaville - BART) averaged 40-50 percent in 2002 and 2003. However, other Vallejo Transit routes achieve similar cost recovery. Vallejo Transit overall achieves strong cost recovery by comparison with other transit systems in the region (VINE route 10 (Vallejo - Napa - Calistoga) averages about 30 percent). It is true that Vallejo Transit route 80 (Vallejo - BART) averages more than 60 percent farebox recovery -- but this line does not parallel the proposed rail line.

Several bus routes from Fairfield/Vacaville and Suisun [City]provide direct service connections to BART in the East Bay, to the Capitol Corridor and to the Vallejo Ferry.

This is another example of incorrect information that has been carried forward from previous study phases. The only current "direct" service to the Vallejo Ferry is Vallejo Transit Route 85, a local route which does not provide automobile-competitive travel times. This rail service could provide much faster connections from Fairfield-Suisun to the Ferry. There are no current connections between any Solano/Napa bus service and the Capitol Corridor rail line that are of any use to passengers traveling between Solano/Napa and points south of the Carquinez Strait.

Minor rerouting of the current Vallejo Transit express Route 80 in downtown Vallejo would provide a useful connection for Solano/Napa rail passengers and BART. About 55 percent of current Route 80 passengers travel via BART to East Bay destinations. Some of these passengers would use the rail service to reach Vallejo as an alternative to driving to Curtola Park & Ride.

[Existing bus routes] are unlikely to be taken out of service.

This statement, and the following "general demographic profile," ignores the fact that rail and bus services are likely to serve two distinct although overlapping market segments.  Some consumers prefer the BART connections at El Cerrito del Norte and Pleasant Hill, while others would use services connecting with the Vallejo Ferry. The Napa/Solano rail service would attract a larger share of the latter than of the former.

A high mode share will not result in high ridership if the market is small whereas a small mode share may create many riders if the market is large.

Having made this important point, the consultant did not present the number of rail commuters for each place listed in Table 6-1 (although this can be derived from "total commuters" and "rail mode share." It is essential to note that Metrolink (southern California) service is skeletal compared to that operated by Caltrain, by MBTA in Boston, and other East Coast commuter rail operators. Metrolink modal shares are lower, but the number of rail commuters is similar. Higher modal shares would require significant expansion of Metrolink service, together with improvements to local transit service and increases in station parking supply.

BART does not attract high mode shares on the strength of its brand name. Of the other rail services listed (ACE, Caltrain, Capitol, Coaster Commuter, Metrolink), only Caltrain provides regular-interval midday service. Nonetheless, BART operates much more service than Caltrain, and so attracts more ridership, as predicted by transit service elasticity models.

The Draft Final Report notes that “commuter rail mode shares are highly variable depending on the ability of commuters to be able to use rail as opposed to driving or other forms of transportation.” As a practical matter, this “ability” is determined to a great degree by the quantity of service operated. Modal share should be compared to level of service operated (e.g. number of trips per day, or annual vehicle-miles per mile of route).

Where a parallel bus route would serve some of the transit market, only half of the potential riders were assumed to be on the train.

This sounds reasonable, but does not reflect actual experience in other markets. A 60 percent rail share is the likely minimum, unless rail service levels are skeletal by comparison to bus. As demonstrated in Los Angeles during the early 1990s, a rail service offering marginally longer end-to-end running times can still attract sufficient patronage from parallel express-bus routes.

Driving between Vallejo and Fairfield is much faster because of the availability of Interstate 80 as well as the more indirect rail corridor route.

This statement, also carried over from previous planning stages, not consider the impact of peak-hour congestion and incidents -- and is increasingly less true during off-peak hours due to growing traffic volumes. Existing express-bus patronage demonstrates a significant base of transit ridership, and this base would be larger if more service was provided.

Ridership would grow by an estimated 163 percent [given 15-minute weekday service] over initial forecasts . . . the effective cost of running many more train sets on a double-track configuration would be substantially greater than the 163 percent of the base project cost, so that the cost per rider would rise.

A 15 minute weekday service would not require double-tracking throughout. This could be accomplished with carefully-located passing sidings at intermediate stations, as has been demonstrated since 1987 by Sacramento’s light rail system, which opened with extensive sections of single track with passing sidings. The operating cost per passenger-mile would be less, and the additional capital and operating expenses would be justified. The marginal expense of operating additional trains is relatively low once overhead, administration, and facility maintenance expenses are covered.

The additional marginal expense of providing midday and weekend service has also not been considered; once train crews are available, it makes economic sense to obtain as much work from them as possible for their eight hours of pay. The additional energy and maintenance costs per vehicle mile for railcars is comparable to buses.

The analysis also did not consider the additional patronage and revenue impacts of providing all day service.

If the approximately 14,000 daily riders projected for 15-minute peak period service constitutes 50 percent of daily patronage – a typical percentage for Bay Area rail transit systems such as BART and San Francisco’s Muni Metro – then total all-day patronage could be 28,000 passengers per day in 2010, and 30,000 daily riders in 2020.

4.  Patronage Forecast for Napa/Solano Rail Transit

Given the shortcomings of the R.L. Banks Reference Rail System as proposed, the authors believe a new patronage forecast was needed, based on more realistic assumptions regarding the nature and extent of proposed rail service, e.g. more in line with 21st Century rail transit practices than the rather archaic, 19th Century-style commuter rail patterns proposed by the Napa/Solano Passenger/Freight Rail Study.

The primary differences between the R.L. Banks Reference Rail System and the conceptual Napa/Solano rail transit network envisaged by the aut hors include:

(1)     Commute period service frequency every 15-20 minutes on both routes, except every 30 minutes north of Napa;

(2)     All-day service every 20-30 minutes;

(3)     Connecting Vallejo Baylink ferry service to San Francisco expanded sufficiently to meet potential demand, plus extensive express bus “bridges” to BART along I-80 and I-680 across the Carquinez Strait;

(4)     “Commuter rail” technical standards to minimize system capital expenses, but also providing sufficient track capacity so that a “transit” level of service can be operated; and

(5)     “State of the art” hybrid battery/diesel-electric locomotives or railcars with acceleration that matches BART railcars. These sorts of propulsion technologies can avoid the high costs of electrification.

The proposed route map is shown in Figure 1 on the next page. The key assumptions for this analysis are frequent service using commuter rail technology and standards, but with a new generation of hybrid diesel-electric railcars that include battery packs to provide “pure electric” acceleration rates, but without the heavy expense of actually electrifying the route. Hybrid technology is now “well-proven” for both automobiles and buses, and ha been successful for railroad switching locomotives. A “commuter rail” design standard is proposed, an attempt to avoid the common problem of some designers to turn new light rail and other rail transit lines into “christmas trees” with unnecessary elaborations and “gold plating.” Capital expenses can be minimized since the proposed rail lines are located completely within existing railroad or road rights-of-way.

The patronage analysis is based on a “direct ridership model approach” developed by Walters and Cervero (2003) for the Bay Area Rapid Transit (BART) District, in a study of a “tBART” (diesel-powered light rail) extension along the I-680 and I-580 corridors between Walnut Creek, the San Ramon Valley, Dublin, Pleasanton, Livermore and Tracy over the Altamont Pass.

According to the abstract of Walters and Cervero (2003):

The ability to justify investments of hundreds of millions of dollars to expand San Francisco Bay Area Rapid Transit (BART) through growing suburban areas depends principally on a demonstration of reasonably good ridership potential. A “direct ridership” model was developed to forecast ridership on alternative light- and heavy-rail BART extensions within a 55-mile corridor. The method is empirically based, derived by establishing statistical relationships between ridership and the characteristics of transit services and surrounding neighborhoods for BART and commuter rail stations in the Bay Area. The method has proven able to:

–address transit alignments, stations locations, and vehicle types

–conduct quick-response evaluation of parking, feeder bus service, and train frequencies

–capture effects of land use density and walkability within station areas and transit-served communities

Combining the direct-demand model with the conventional 4-step travel models accounts for both macro travel patterns and micro-area sensitivities for multi-station transit concepts. Relying upon empirical experiences from the same metropolitan area also enables policy relevant sensitivity tests. For example, transit oriented development was found to boost ridership by 11% to 17%. Verification against ridership on comparable transit systems and corridors demonstrates that the forecasting method produces reasonable results and is sensitive to different assumed transit and land use futures.

In an accompanying document, tBART 580/680 Corridor Ridership Forecasting Methodology, (2003), development the direct ridership model is described:

The formula used to forecast ridership at new tBART stations was derived from statistical analysis of independent variables related to year 2000 peak period boardings and alighting counts at all BART and [Caltrain] stations. It also developed data on over 30 prospective independent variables believed to potentially be correlated with station ridership. We tested the relationship of ridership to the following variables both individually and in combination:

The list of variables is presented in Table 4.

Table 4. Variables Tested in tBART Extension Analysis

Variable

Population and employment within 1/4 mile of station

Population and employment within 1/2 mile of station

Population and employment within 1 mile of station

Catchment-area population

Rolling-average catchment population for station and nearest neighbors

Mean travel time to all other transit stations, weighted by relative attractiveness of each station

Destination-weighted transit fare

Employment density, surrogate parking cost

Household income

Parking spaces

Terminus station

Freeway-intercept station

Vehicle type

System speed

Train frequency

Station spacing

The discussion continues:

The statistical analysis including factor analysis and linear regression and log/log regressiona analyis of combinations of variables to discover the combinations of variables with the strongest statistical correlation with ridership. The two formulae found to have the highest correlation (R-squared) and include a combination of system attributes and land use and demographic characteristics are the following:

(1)  RIDERSHIP =

    2.04 + 0.300 X POPEMP + 0.069 X POPCTCH + 0.560 X TRAINS + 1.787 X TECH

(2)  RIDERSHIP =

2.400 + 0.233 X POPEMP + 0.021 X POPCTCH + 0.287 X BUS + 0.038  X PARK + 0.477 X TRAINS + 1.576 X TECH

Where:

    RIDERSHIP = Station ridership

    POPEMP = Population and employment within 1/2 mile of station

    POPCTCH = Catchment-area population

    BUS = Feeder bus service level (buses per hour)

    PARK = Parking spaces [at station]

    TRAINS = Train frequency

    TECH = Train vehicle type (BART = 1, Caltrain - 0)

Note: The log transformation performed on all of the variables is LN(X+1).

The R-squared for this formulation is 0.87, indicating that the formula explains 87% of the variation in ridership among transit stations on the existing BARt and Caltrain system.

In addition, the empirical analysis and detailed sensitivity testing to the ACCMA and CCTA models revealed the following elasticities relating ridership to individual system attributes:

This data is presented in Table 5.

Table 5.  tBART System Atttributes Patronage Elasticity

Attribute

Elasticity

tBART System Speed

0.41

Core BART System Speed

0.20

tBART Service Frequency

0.35

Core BART Service Frequency

0.04

Station spacing

0.14

Station bus access frequencies

0.29

TOD design

0.04

TOD Density concentration within 1/4 mile

0.05

Formula (2) above was chosen for the patronage forecasting exercise documented in this report due to inclusion of feeder bus service levels and other factors known to the author. The estimate of daily passenger miles generated by each station and segment was based on the authors’ knowledge of station locations and distances between major destinations. For detailed documentation of these patronage calculations, please contact the authors.

It was also necessary to make several assumptions about potential ridership by visitors to Napa and Solano Counties, particularly along the route between the Vallejo Ferry Terminal and the Napa Valley. For this purpose, estimated annual visits to particular subareas of the Napa Valley, specifically downtown Napa, Yountville, Oakville/Rutherford/Zinfandel, St. Helena, and Calistoga were estimated, doubled to obtain “trip ends” and 10 percent of these “trip ends” estimated to be boardings and alightings by visitors. This percentage appeared to be reasonable, since Napa Valley visitor trips are concentrated in the listed subareas, and an all-day, frequent service is proposed. Currently, the Napa Valley Wine Train attracts about 3 percent of annual Napa Valley non-business visitors with only two daily trains.

Figure 1.  Potential Napa & Solano Rail Transit

NapaSolanoRailRoutes2.pdf

The results of the authors’ analysis is summarized in Table 6.

Table 6.  Summary of Napa/Solano Rail Transit Patronage Projection, 2010 and 2020

Route Segment

Peak Headway

Total Ons & Offs 1

Total Daily

Passenger Miles 2

Year

 

2010

2020

2010

2020

Calistoga – St. Helena

30 min.

1,472

1,476

30,916

30,990

St. Helena – North Napa

30 min.

3,492

3,774

51,566

55,796

North Napa – Vallejo Ferry Terminal

15 min.

31,177

32,557

53,759

57,705

Vacaville – Northeast Fairfield

15 min.

5,913

6,061

118,305

121,335

Northeast Fairfield – American Canyon

15 min.

9,242

9,563

145,504

150,535

TOTAL Ons & Offs

 

51,296

53,431

400,050

435,361

DAILY TRIPS 3 / Average Trip Length

 

25,648

26,715

~16.0

~16.0

Traffic Density (Pass mi / route mi)

     

6,700

7,012

1  Includes visitor trips

2  Based on estimated trip length for each station.

3  Total ons and offs divided by two

Table 7 summarizes projected patronage by proposed station.

Table 7. Projected Station Volumes, 2020 Data

Station

Mileage  Vallejo Ferry

Daily ONs & OFFs

Visitor ONs & OFFs

TOTAL

Incre- Pass. Miles

Incremental Traf Den

Davis Street

27.5

3,042

0

3,042

   

CA Medical Facility

25.5

3,019

0

3,019

   

Vacaville Segment

 

6,061

0

6,061

121,335

24,267

NE Fairfield/Travis AFB

22.5

958

0

958

14,370

 

Sunset Avenue

20.0

2,599

0

2,599

38,935

 

Suisun City Amtrak

18.5

3,545

0

3,545

60,265

 

Solano Business Park

17.0

1,129

0

1,129

16,935

 

Cordelia

12.5

680

0

680

10,200

 

Red Top Road

11.5

652

0

652

9,780

 

Fairfield Segment

 

9,563

0

9,563

150,535

12,545

Calistoga

42.0

578

438

1,016

22,510

 

Dunaweal Lane

38.5

350

110

460

8,480

 

Calistoga Extension

 

928

548

1,476

32,006

3,443

St. Helena / Sutter Home / Zinfandel

32.0

870

1,370

2,240

33,600

 

Rutherford/Oakville

28.0

454

548

1,002

14,216

 

Yountville

25.0

532

0

532

7,980

 

St. Helena Segment

 

1,856

1,918

3,774

55,796

3,985

Oak Knoll / Wine Country Avenue

18.5

690

0

690

8,970

 

Redwood Road

17.0

3,767

0

3,767

37,670

 

California Boulevard

16.5

800

0

800

8,000

 

Lincoln Avenue

15.5

840

0

840

8,400

 

Napa CBD

14.5

3,641

2,341

5,963

68,918

 

Imola Avenue

13.5

1,018

0

1,018

10,180

 

Napa College

13.0

657

0

657

5,256

 

Napa Airport

9.0

973

0

973

9,730

 

Napa Segment

 

12,386

2,341

14,747

157,124

10,474

American Canyon / Amer. Can. Rd.

5.0

830

0

830

3,320

 

Mini Drive

4.3

2,158

0

2,158

12,948

 

Marine World Parkway

3.5

723

0

723

4,338

 

Sereno Transit Center

2.5

1,514

0

1,514

7,570

 

Sonoma Blvd/Nebraska

1.5

745

0

745

3,725

 

Mare Island Way / Mare Island

0.8

1,448

0

1,448

7,240

 

Vallejo Ferry Terminal

0.0

10,431

0

10,431

156,465

 

American Canyon / Vallejo

 

17,849

0

17,849

352,730

9,798

TOTAL

 

48,643

4,807

53,480

869,526

 

TOTAL Daily Trips & Pass. Miles

 

24,321

2,404

26,725

434,763

7,012

Patronage predicted in this analysis is summarized and compared to the projections developed for the R.L. Banks Reference Rail System in Table 8.

Table 8.  Publictransit.us Analysis vs. R.L. Banks Reference Rail System

WEEKDAY VOLUMES

2010

2010

2020

2020

Boardings & Alightings *

R.L. Banks

This Analysis

R.L. Banks

This Analysis

Vacaville – Fairfeld/Suisun City

0

5,913

0

6,061

Fairfield/Suisun City – Vallejo

3,464

9,242

3,868

9,564

Calistoga – St. Helena

0

1,472

0

1,476

Vallejo – Napa – St. Helena

4,732

34,669

5,412

36,331

Fairfield/Suisun City – Napa

3,788

0

4,124

0

TOTAL SYSTEM

12,324

51,296

13,404

53,432

Total Passenger Trips

6,162

25,648

6,702

26,716

* e.g., ons and offs

In general, the level of Napa/Solano rail transit ridership estimated in this report would quadruple total patronage over the R.L. Banks estimates. This is due mainly to the much higher level of service and proposed extension of service to Vacaville and Calistoga.

The estimates in this report are also consistent with the R.L. Banks “sensitivity analysis” conducted to evaluate the impacts of operating trains every 15 minutes rather than every 60 minutes, as proposed for the R.L. Banks Reference Rail System. The R.L. Banks sensitivity analysis for 15-minute peak period service projected a total of about 14,600 and 15,700 daily passengers in 2010 and 2020, respectively. The Publictransit.us analysis is actually very close to the R.L. Banks estimate if it is assumed that off-peak rail patronage would be about one-third of the daily total, as illustrated in Table 9 below, and considering additional ridership attracted by the proposed Vacaville and Calistoga extensions.

Table 9.  “All Day” Service Scenarios Compared

Scenario

2010 Projection

2020 Projection

R.L. Banks, PEAK ONLY

14,600

15,700

R.L. Banks, ALL DAY (+50%)

21,900

23,550

Publictransit.us ALL DAY

25,600

26,700

5.  Discussion

The term “traffic density” describes the number of passengers who, on average, travel over each kilometer (or mile) of system length. The average trip length per passenger for the Napa-Solano rail scenario studied here is projected to be 15.6 miles in both 2010 and 2020 . On a typical weekday, 25,600 and 26,700 passengers would travel about 4 2 0,000 and 4 35 ,000 miles, respectively in 2010 and 2020.

The system’s overall passenger “traffic density” is projected to be about 6,500 and 7,000 “daily passenger miles per route mile” (dpm/rm) in 2010 and 2020, respectively. federally-funded research from 1949 and the early 1980’s, and independent Japanese and European studies, indicate that as a general rule, the minimum “threshold” of feasibility for rail transit is a traffic density of 5,000 dpm/rm. BART carries about 40,000 dpm/rm, and Sacramento light rail, 7,700 dpm/rm. For the Sonoma/Marin "SMART" commuter rail project, the authors project 2,500-3,000 dpm/rm.

Research by Pushkarev, Zupan and Cumella (1980) also estimated that initial transit capital expenditures of $1,250 per daily passenger mile per route mile (19 77 dollars ; approximately $ 4 ,000 in 2005 dollars) were justified based on estimated travel and other benefits. Based on this yardstick, system capital expenditures of up to $1.7 billion could theoretically be justified for a Napa/Solano rail transit network such as that hypothesized in this report. “Commuter rail” design standards could hold this estimated expense to $400-$500 million. A rail extension “Up Valley” north of Napa would not be justified using the “traffic density” threshold. However, many passengers would be tourists, who could be charged higher fares to offset this projected traffic density shortfall.

This analysis also suggests that a rail transit connection between BART in Contra Costa County and the Vallejo Ferry Terminal would be very productive, and could generate passenger traffic density in the range of 15,000–20,000 daily passenger miles per route mile.

References

Brinkman, P. Anthony. 2003. The Ethical Challenges and Professional Responses of Travel Demand Forecasters. Doctoral dissertation (City and Regional Planning, University of California, Berkeley). http://www.uctc.net/scripts/countdown.pl?diss106.pdf

Demery, Leroy W., Jr. 1994. “Supply-Side Analysis and Verification of Ridership Forecasts for Mass Transit Capital Projects.” Journal of the American Planning Association  60, 3: 355-371.

Demery, Leroy W., Jr. 2005. Supply Side Analysis & Verification of Ridership Forecasts: A Retrospective. http://www.publictransit.us/ptlibrary/specialreports/sr4.supplysideretrospective.htm

Demery, Leroy W., Jr. 1997. An Analysis of Ridership Forecasts for the Los Angeles Metro Red Line: Alternative Strategies and Future Tshapeimage_4.pngransit Improvements (publictransit.us Historic Monograph MG 02–03). http://www.publictransit.us/ptlibrary/specialreports/LosAngelesSubway.htm

Demery, Leroy W., Jr., and J. Wallace Higgins. Rapid Bus and Rapid Rail: Peak Period Service Supply vs. Observed Passenger Utilization. 2003. http://www.publictransit.us/ptlibrary/modalcapacity5.pdf

Demery, Leroy W., Jr., and J. Wallace Higgins. Rapid Bus and Rapid Rail: Peak Period Service Supply vs. Observed Passenger Utilization. 2003 . (Abridged version of paper above, presented at 9th National Light Rail Transit Conference, Portland, OR, November 16-18, 2003.) http://gulliver.trb.org/publications/circulars/ec058/09_CRITIQUES--HOW%20ARE%20WE%20DOING.pdf

Demery, Leroy W., Jr. 2005. Peak-Period Vehicle Occupancy Statistics for U.S. and Canadian Rapid Bus and Rapid Rail Services . http://www.publictransit.us/ptlibrary/peakoccupancy41805.pdf

Demery, Leroy W., Jr., and Michael D. Setty. Ridership Forecasts and Traffic Density. 2004. http://www.publictransit.us/PublicTransitUS.php?itemid=126&catid=11

Demery, Leroy W., Jr., J. Wallace Higgins and Michael D. Setty. 2005. Traffic Density Thresholds for Rail Transit: A Retrospective (publictransit.us Special Report No. 2). http://www.publictransit.us/ptlibrary/specialreports/sr2.trafficdensityretrospective.htm

Fehr & Peers Associates, Transportation Consultants (Gerald Walters). 2003. tBART 580/680 Ridership Forecasting Methodology. June 2003. Lafayette, CA. Prepared for the San Francisco Bay Area Rapid Transit (BART) District.

Litman, Todd. 2005. Rail Transit in America: A Comprehensive Evaluation of Benefits. Victoria Transport Policy Institute, Victoria, British Columbia. www.vtpi.org/railben.pdf .

Mokhtarian, Patricia L., Salomon, Ilan, Redmond, Lothlorien S. 2001. Understanding the Demand for Travel: It’s Not Purely Derived.” Innovation: The European Journal of Social Science Research 14(4), 355-380, 2001. (Paper originally prepared for the THINK-UP Workshop on Policy Factors and Mobility Prediction: Transport Planning and Modeling, sponsored by the Interdisciplinary Centre for Comparative Research in the Social Sciences, Vienna, Austria, May 21-22, 2001).

Napa County Transportation Planning Agency and Solano Transportation Authority (2003). Napa/Solano Passenger/Freight Rail Study. Developed by R.L. Banks & Associates, Inc. and associated subcontractors.

Ory, David T. and Mokhtarian, Patricia L. 2005. “When is Getting There Half the Fun? Modeling the Liking for Travel.” Transportation Research Part A 39 (2005) 97-123.

Pickrell, Donald H. 1989. Urban Rail Transit Projects: Forecast versus Actual Ridership and Costs. Cambridge, MA: Transportation Systems Center, U.S. Department of Transportation.  (Prepared for Office of Grants Management, Urban Mass Transportation Administration, U.S. Department of Transportation.)

Redmond, Lothlorien S., and Mokhtarian, Patricia L. 2001. “The Positive Utility of the Commute: Modeling Ideal Commute Time and Relative Desired Commute Amount.” Transportation 28: 179-205, 2001.

Tennyson, E. L. 1989. “Impact on Transit Patronage of Cessation or Inauguration of Rail Service.” In Transportation Research Record 1221. Washington, DC: Transportation Research Board, National Research Council.

Online at www.publictransit.us/ptlibrary/supplydemand/TRB1221.pdf

U.S. Department of Transportation, Urban Mass Transportation Administration and Southern California Rapid Transit District. 1983. Draft Environmental Impact Statement and Environmental Impact Report, Los Angeles Rail Rapid Transit Project.

Walters, Gerald, and Cervero, Robert. 2003. Forecasting Transit Demand in a Fast Growing Corridor: The Direct-Ridership Approach. Fehr & Peers Associates, Lafayette, CA., Department of City and Regional Planning, University of California, Berkeley. August 2003.