Rail Draft Environmental Impact Statement: Technical Comments

My comments on the City and Count of Honolulu's Rail Draft Environmental Impact Statement (DEIS) were reorganized into two posts. The previous post covered general concerns, and this post covers my technical concerns of the report.

My review was based upon the DEIS section 4F dated November 2008 and particularly of chapter three on transportation impacts. Many of my comments refer to the supplementary report “Transportation Technical Report, Honolulu High-Capacity Transit Corridor Project, Prepared for: City and County of Honolulu, 417 pp, August 15, 2008” which includes much more detail and explanations on the traffic and transportation analyses that were the foundation of the results presented in the DEIS.

• Traffic Analysis Methodology

The traffic analysis method used is not suitable for saturated conditions, and is not suitable for corridor and regional studies. HCM mentions these limitations. Almost all traffic elements along this corridor are oversaturated, thus HCM methodologies do not apply (unless the wrong data are used and degrees of saturation are low.) Either way the output is wrong or misleading.

The table below, in which all black cells are the reviewer’s corrections, shows that general purpose traffic was estimated to be 31% above capacity (estimate of 1.31) but by their numbers, the correct estimate is 62.5% over capacity (estimate of 1.625.) Capacities are not revealed everywhere in the DEIS, so the reviewer cannot check the same calculations in the DEIS.

• Forecasts

Neither the DBEDT (provider of some of the base forecasts), nor the City nor their consultants understand that most growth phenomena in a metropolitan area concerning city expansion and their traffic follow an S-curve depicted by many years of existence as a village, transitioning to a city, several years of growth into a metropolitan area followed by a very long period of maturity with small growth (and decrease) periods. This study erroneously assumes a large future growth for west Oahu and nightmare traffic scenarios whereas Oahu's population, development and tourist attraction have ended their sharp growth and have entered their mature level with a lot of negative bumps along the way. For example, DBEDT Data Book Table 1.06, Honolulu population in 2006 was 906,715 and it dropped to 905,601 in 2007 which was before the sharp economic downturn of late 2008 which is expected to last till 2011.

As shown above, if S-shape forecasts were used, then the unrealistic demand levels shown in the Alternatives Analysis (AA) would never had appeared. However, something inexplicable happened between AA and DEIS: Screenline demands have been reduced by 28% without any explanation. As shown in the table on page 3, demands in the 2008 DEIS are lower by 28% for year 2030 compared to what they were in the 2006 Alternatives Analysis.

Such a discrepancy (28%) in demand produced by the OMPO forecasting model is highly suspect. Qualified alternatives such as TSM and Managed Lanes were dismissed based on high demand figures in the AA which were subsequently modified in the DEIS. A supplemental DEIS is needed to evaluate qualified alternatives with the reduced demand forecasts.

• Were ORTP 2030 Congestion Relief Projects Modeled Correctly?
  

Page 3-16: "Even with $3 billion in roadway improvements under the No Build Alternative, traffic delay in 2030 would increase by 44%".

If one was to correctly model all the committed congestion relief projects in ORTP 2030 (Table 2-3) and combine them with a the fact that Oahu population has been stagnant or falling (and bound to further fall due to poor economy and housing unaffordability), the highway congestion in 2030 could be improve by at least 15%.

For example, the PM zipper alone will carry about 1,500 vph through the Kalauao screenline with 3 or more people in them resulting in a person capacity of 4,500 going west. These are people removed from the existing network thus providing a substantial relief.

The westbound utilization of the rail will be optimistically 6,000 people through the Kalauao screenline of whom at most half will be drivers and ex-carpoolers or 3,000 people.

The PM zipper combined with a Nimitz flyover practically guarantee a continuous trip at 55 mph from Iwilei to Waikele to Kapolei. This commute is half as long in duration as that by rail.

Therefore, the PM zipper alone that carries more persons than rail can be more beneficial that rail. However, the DEIS tries to convince us that major traffic congestion relief projects will yield “peanuts” whereas the rail with its inferior speed and 15+ stops to Kapolei will yield superior travel time savings and traffic congestion improvements.

Part of the reason is likely that planning models are insensitive to bottlenecks and only provide rough estimates based on some assumed values of capacity. Until this author sees proof of use of a regional microsimulation traffic model assessing the impacts without and with correctly modeled ORTP 2030 projects, he asserts that the analysis method was inappropriate and largely incapable in assessing the benefit of the projects in Table 2-3 of the DEIS.

• DEIS Base Travel Times Are Inaccurate

Having resided in Kapolei for a short period if 2007, I know from personal experience that the morning peak period travel time from Kapolei to downtown is always under 75 minutes in the absence of rain or any lane closure. I was startled that the DEIS uses a time of 89 minutes.

I took the opportunity to ask people listening to a radio program that I participate to make some measurements of travel time from the H-1 freeway on-ramp to Alakea Street in downtown if they depart Kapolei between 6 AM and 7 AM. So far I received six qualified measurements of 49, 62, 75, 50, 62 and 59 minutes averaging at about 60 minutes. Therefore, roughly speaking the DEIS uses a 50% overestimate of the travel time which leads to false benefits of travel times by rail.

The DEIS fails to demonstrate the root causes of traffic congestion. The same travelers reported these airport-to-Alakea travel times: 18, 16, 41, 11, 30 and 25 minutes for an average of 23.5 minutes (DEIS uses 25 minutes). The real issue therefore is the traffic flow condition on Nimitz Hwy. which vary widely as these travel times show: 11, 16 or 18 minutes with good conditions, 25, 30 or 41 minutes with poor conditions. This makes it clear that a roughly two mile long Nimitz Viaduct will provide a consistent travel time from airport-to-Alakea of about 6 minutes, reducing the peak hour trip from Kapolei to downtown from about 60 minutes to about 40 minutes. A relatively modest investment solves a huge part of the morning commute congestion.

Note that rail will be providing airport-to-Alakea transit travel time of about 50 minutes (It is 50 to 54 minutes depending on the route selected. The airport route provides the longest travel time for this origin-destination pair.)

• TheBoat as a Threat to the Rail

TheBoat vessel inventory in page 3-31 is wrong. It should also be mentioned that its schedule reliability is poor due to frequent mechanical failures and high seas.

Since we spend the significant amount of $6 million a year on TheBoat, why didn't the DEIS estimate the productivity and congestion reduction of this alternative transportation mode? Will TheBoat reduce rail's ridership?

• Forecasts from the OMPO model

There is a long list of limitations of the OMPO model used to develop the all-important rail forecasts. Here are a few:

1. The model was developed in 1994 by Parsons Brinkerhoff. It is very old in terms of both architecture and data validity. It is also of interest that the same person who developed it as a Parsons Brinkerhoff forecaster now is an Federal Transit Agent who inspects the forecasts.
2. The model has parameters for dead attractions such as the Kodac Hula Show and the Dole Cannery, but has not parameters for Superferry, Ko Olina, Water Adventures Park, North Shore and Haleiwa.
3. The OMPO model is hardly a modern activity-based microsimulation platform. It is an old, aggregate platform with highly compartmentalized trip definitions.
4. The OMPO model depends on many assumed static capacities for various facilities. This makes it susceptible to range errors and easy manipulations. Note that the transit factor table depends on congested times. It would make sense that more people would choose transit from Kapolei to downtown if a time of 90 minutes is used instead of the correct time of 60 minutes. And that was done.

Same concern applies to arterial and freeway capacity which can be arbitrarily set too high or too low to satisfy the objective of the analysis such as “promote rail and undercut HOT lanes.”L