MTC study says that adding rail
is comparable to building a
whole new bridge

Errors and hundred-fold increases:
our critique of MTC's flawed rail study

The Metropolitan Transportation Commission's (MTC's) Bay Bridge rail feasibility study was the subject of an hour-long presentation at the offices of the San Francisco Planning and Urban Research Association (SPUR) on February 16, 2000. We had earlier questioned an extraordinarily large weight loading value (dead load) in the design of a proposed West-Span rail deck, a number which appeared in the printed draft version of the study. The engineer presenting the structural assessment phase of the study replied that the number was about 2.5 times what it should have been, but that fortunately (?) it was a typographical error. He added that the error had not affected other numbers in the study, hence the total loading of the bridge with rail deck and train added (he said) remained a very high number. We disputed this because, while we could not prove that the typographical error was solely responsible for the inordinate load numbers, there was something which did not add up to the consultant's conclusion that an added rail deck would require extensive modifications to the towers and an additional set of main cables! The West Span, which once carried trucks and trains on its massive lower deck, has a very large reserve load capacity, and that reserve capacity will actually increase whenever Caltrans completes its planned replacement of the automobile decking with modern lightweight materials.

Nevertheless (and evermore), MTC's high estimates were publicized. Rail on the Bay Bridge, found to be technically feasible, was also found to be economical—until cost and weight estimates for many design elements were inflated by factors as high as 100. The inflated estimates appeared shortly after a committee of technical advisors was consulted about design and structural issues. Each member of the technical advisory committee—save the one who did not participate—was either an employee of Caltrans engaged in designing the proposed east-span replacement, or was a member of the design team contractually committed to the success of the proposed design, or was a consultant with a biased and confidential relationship to Caltrans. Were the study's faults and errors only typographical? Based partly on acknowledged errors, an estimate of less than $1 billion was inflated to over $3 billion. By strange coincidence this is the same number MTC projected as the cost of rail on the bridge before the study was begun.

Some transit advocates actually said that rail should be put on the bridge at any cost. This must have panicked the MTC, since they tacked on another $5 billion, then announced the conclusion of their rail study (July, 2000) with yet another headline-grabbing number—$8 billion—which is what, they said, it could cost to equip the Bay Bridge with rail. Then MTC moved on to study (endorse, we should say) building a multi-billion-dollar bridge in addition to the Bay Bridge—a "southern crossing"—using California Senator Feinstein's support and the results of its Bay Bridge rail infeasibility study to justify opening up new territory for its consultants to explore.

[January 21, 2000]

by Daniel Coman, Rick Feher and Robert Freehling

A study is currently underway to assess the feasibility of restoring passenger rail service on the San Francisco-Oakland Bay Bridge. The following essay shows that gross errors, particularly the mistaken addition of about 80 million pounds to the bridge’s dead load, render the study’s preliminary findings to be invalid. Our analysis began when we discovered the almost unbelievable errors discussed below. The Metropolitan Transportation Commission (MTC), obliged by a public initiative to study rail on the bridge, has used the study’s findings to dismiss the idea of re-establishing rail on the Bay Bridge, even though the study’s initial findings found rail to be feasible and economical. Looking further into the matter, we discovered the MTC may have an insidious agenda.


In November, 1998, citizens of four Bay Area cities voted overwhelmingly to advise planners of a proposed new San Francisco-Oakland Bay Bridge eastern span to consider putting passenger trains back on the bridge. (Passenger rail service existed on the bridge from 1937 through 1958; at one time most people crossed by rail.) While a non-standard local service, Bay Area Rapid Transit (BART), serves trans-bay commuters locally, travelers entering San Francisco from other cities, or San Franciscans going elsewhere, have been unable to get from here to there except by automobile, or by negotiating journeys involving two or more of the following: train, bus, automobile, bicycle, walking, ferry, BART. The Bay Bridge is the most heavily traveled long-span bridge in the world, forming the principal connections to California’s main transportation corridor. Restoring a standard rail link to San Francisco and its peninsula is a top priority in restoring rail service statewide.

The rail initiative was only ‘‘advisory’’—backers of the new span had legislated that the hurried project be exempt from state environmental review and from any requirement that a mass transit facility (i.e., trains) be included before construction would begin. The California project could not be exempted from federal statutes, however, which require that rail be given serious consideration. Project coordinators were also notified by the U.S. Environmental Protection Agency that rail feasibility needs to be determined.

The popular opinion expressed in the rail initiative was that, since the replacement span would be a major capital project, it ought to maximize public transit. That idea was specifically rejected by the Metropolitan Transportation Commission (MTC), and the commission voted instead to approve a bridge type which, at 30 percent completion of the design, was already locked into its destiny as an elevated freeway with no mass-transit options.

In response to the rail initiative, MTC commissioned a feasibility study to assess the following aspects of restoring train service to the Bay Bridge:

Structural modifications necessary to accommodate rail on the existing west span, the proposed east span, on Yerba Buena Island, and for transitions at both ends;
Rail service type, routing and operational options;
Patronage estimates and capital and operating cost estimates for all feasible options.

According to MTC, the rail study was ‘‘scheduled to wrap up in December 1999.’’

The structural assessment phase was released to the public as Working Paper 2A.3: Structural Assessment of Rail on the Bay Bridge (W.P. 2A.3) on October 22, 1999. According to the title page, this assessment was produced by a group of consulting firms: Nelson/Nygaard Consulting Associates; Ove Arup & Partners; LTK Engineering Services; Simon Martin-Vegue Winkelstein Moris; ‘‘and Associated Consultants.’’

‘‘Structurally there are ways of modifying the Bay Bridge to accommodate rail,’’ concludes W.P. 2A.3 (XI). Since the Bay Bridge formerly supported rail, any other conclusion would sound strange. The purpose of the structural phase of the study, it follows, would be to weigh advantages and disadvantages of design alternatives, and to determine costs. But the costs detailed in W.P. 2A.3 are based on questionable assumptions and massive quantitative errors. Each of these—the assumptions and the quantitative errors—affects the other, causing the whole study to derail.

Incredibly, a new third deck for rail proposed in W.P. 2A.3 features primitive 1930s design. One wonders whether the authors were intent on mimicking the style of the existing bridge—which employs 19th-century technology— instead of using more efficient materials and techniques common in the second half of the 20th century. Worse, the study contains an error of calculation so conspicuously large that it at first appears to be a typographical error. But this theory collapses when one sees the error reflected in every table and chart in the study. The study appears to demonstrate that rail would weigh so heavily on the bridge that almost every part of it would have to be reconstructed, at enormous cost.


Ove Arup proposed a scheme for a train deck suspended under the double deck of the western suspension spans (W.P. 2A.3, pp. 17 and 27). The proposed structure is extremely heavy (likely the heaviest possible) consisting of two angled Warren trusses connected by the rail platform. The platform consists of primary beams, secondary beams and an eight-inch-thick concrete slab. The two angled trusses are shown as hinged at the bottom of the existing stiffening trusses of the suspension spans, their connections to the primary beams supporting the train platform being rigid (moment) connections. Being moment connections, they must resist all lateral forces. Such forces (induced by earthquake or wind) depend upon, and are proportional to, the total mass of the structure. Note that:

moment connections—extremely inefficient in both material and performance—require a great deal of mass to maintain rigidity;

the proposed multiple-beam design is nearly identical to the 1930s deck which supported heavy rail and trucks;

an eight-inch-thick concrete slab would further increase mass, requiring, in turn, still more mass in the superstructure. (A less massive rail-deck substrate could serve an identical function.)

The point at which there is enough mass to resist possible lateral forces brings the total structural weight to a very high number. (See ‘‘Numbers,’’ below.)

On page 29 (figure 17) of the same document is a cross section of the proposed east-span decks depicting a steel box-girder design. In their modern iterations these ‘‘orthotropic’’ steel designs are considerably lighter than multiple-beam designs. Why then has Ove Arup elected to propose an extremely heavy design for the train deck under the western span, when the projected weight increase, according to the paper, would require additional main suspension cables and additional strengthening of the suspension towers and foundations, all at enormous cost?

John Eddy of Ove Arup, who publicly presented results of the structural assessment, admitted to us that a lightweight design could have been proposed. He offered the excuse that, since this was a ‘‘feasibility study,’’ he felt it was not feasible to try to have such a design approved by a certain peer-review panel (the Technical Advisory Committee) prior to the document’s release. Curious, because the Technical Advisory Committee is comprised of several key individuals who designed and/or approved lightweight steel decks for the proposed eastern span which are depicted on page 29. Even more astounding: lightweight ‘‘orthotropic’’ steel decks are already approved as part of an upcoming extensive retrofit and upgrading of the west spans.

Clearly, the object of the proposed design should have been to assess whether a train platform can be built to current seismic specifications at a reasonable cost and with the least impact upon the existing structure.

In W.P. 2A.3, the exact opposite was accomplished. Between Ove Arup’s initial findings and peer-review alterations, some items were inflated by over 10,000 (ten thousand) percent. (Compare items in Ove Arup’s first estimates with items in W.P. 2A.3, Appendix A.) Among these items is a tube through the West Span’s center anchorage which is, at that elevation, an essentially hollow masonry structure 150 feet long. Initially estimated at $1 million, the tube gains two zeros to become $100 million in the second set of figures.

Reduction in vertical clearance over shipping lanes for the below-deck design of the west spans is given as 32 feet. This reduced clearance is cited as a potential problem, subject to approval by the Coast Guard. A conventional passenger-train envelope, including allowances for overhead equipment and structural sizing, would reduce this figure to 22 feet.
Figure 15, page 27 is an illustration of the below-deck concept, appearing enormously massive and overbuilt, in contrast to the graceful proportions of the bridge as it appears today. If anyone wanted to object to a train deck on aesthetic grounds, this illustration could be used to offend aesthetic sensibilities—and might even generate protests—from architects and others for whom the Bay Bridge is an inspiring landmark.

Bonnie Nelson of Nelson/Nygaard said, at a public meeting on December 17, that the structural assessment (the topic of W.P. 2A.3, including cost estimates) was a completed phase of MTC’s rail study.

Numbers: feet, pounds and dollars

Insofar as the strength of the engineering profession may not be its capacity to bring forth good design, the antiquated and inappropriate design solution proposed by Ove Arup for putting rail on the western suspension spans may be forgiven, but not accepted. The strength of the engineering profession is presumed to be in its ability to find truth in numbers.

Here this does not seem to be the case.

The drawing on page 17 of W.P. 2A.3 shows a dimension of eight inches (8") for the concrete slab of the train deck. On page 21 (table 6), however, is found an incredible number: 11,000 pounds per each linear foot of the 34-foot wide deck. If delighted cement contractors actually succeeded in pouring 11,000 pounds of concrete per linear foot onto the rail deck, that slab would be 26 inches thick, not eight inches, as shown in the drawing.

John Eddy, whose public presentation of the findings of W.P. 2A.3 took place October 22 in San Francisco, admitted to us at the following rail study meeting, December 17, that the most preposterously wrong calculation in the study — 11,000 pounds per linear foot for a concrete slab — was a mistake.

Unfortunately (or—perhaps—intentionally?) this enormous mistaken weight, equal to the weight of nine (9) trains at 900 pounds per foot of train, was carried through in all of the comparative graphs and tables in the entire paper. Of course those graphs and tables are, as a result, worthless.

Table 6, page 21 of W.P. 2A.3 contains additional erroneous or inflated numbers:

Rails, fasteners, plinths, power and control systems (Item 2) are shown to weigh 600 pounds per linear foot of bridge. In reality, 132-pound rail (the heaviest used on any main line in California) is defined as weighing 132 pounds per yard. On a two-track system, meaning four strands of rail, this equals 176 pounds per linear foot. Rail experts tell us that fasteners, wire, etc. cannot possibly weigh more than an additional 40 pounds per linear foot, when standard lightweight materials are used. Therefore the above item, at 600 pounds per linear foot, is about three times as heavy as today’s industry standards.

On the same table, the first item—steel superstructure—weighs in at 5,920 pounds per linear foot of bridge. As previously stated, this is a very heavy structure. A moderately light orthotropic steel superstructure would weigh close to 3,500 pounds per linear foot, including two six-foot bike lanes.

Considering that, according to W.P. 2A.3 page 24, table 8, the scheduled replacement of the existing bridge’s two automobile decks (orthotropic steel deck) would yield weight savings of 4,060 pounds per linear foot, and that passenger trains weigh 1,800 pounds per linear foot (both parallel tracks loaded), the total weight deficit is reduced to about 1,456 pounds per linear foot. This added weight is easily within the design capacity of the current bridge: section IV of the working paper states, “a 10% increase in total load would likely be tolerated for service load conditions.”

W.P. 2A.3

Our corrected
lightweight steel superstructure
Rail superstructure
Concrete deck
Rail & equipment
Rail live load range
Main bridge strengthening
Second seismic retrofit
Total, including heaviest live load
Savings with 4,060- lbs/lf orthotropic steel deck 16,860 8,860 1,456


However, prefabricated aluminum decks, available for over forty years and now routinely used in new, replacement and retrofitted bridges, would provide comparable performance and would yield weight savings of 6,090 pounds per linear foot of bridge.

Intelligent design using standard materials and procedures would yield a net weight savings of 790 pounds per linear foot (6,090 lbs. less 5,500 lbs. for train deck and trains). Such a solution would obviate the need for additional cables, strengthening of towers and their foundations, and other attendant strengthening of the superstructure.

Due to compounded errors and Titanic structural excesses, W.P. 2A.3 proclaims a net weight increase of 16,860 pounds per linear foot of train deck (including train), necessitating, in turn, extremely expensive cable, tower and foundation work.

The conclusion of W.P. 2A.3 (IX) is, therefore, that retrofitting the western and eastern spans for rail would cost more than three billion dollars. Using this outrageous figure, MTC concludes:

In the context of structural feasibility only, rail on the Bay Bridge is possible.
[ ¶ ]However, restoring rail to the Bay Bridge is very costly based on this ‘first-cut’ study. [ . . . ¶ ] It should also be noted that adding rail to a structure (west spans) that is more than 60 years old complicates the project and reduces the ultimate life of the entire project. Therefore, alternate means of crossing the Bay with rail should be evaluated and compared with adding rail to the Bay Bridge.

The foundation of MTC’s pessimistic conclusion is W.P. 2A.3 with its proposed overweight design and gross errors of calculation which in turn impact the design assumptions. The inordinately inflated cost estimates of W.P.2A.3 are a direct result of this. The $3 billion cost estimate is more than $2 billion higher than the consultant’s own first estimates, making it appear as if only modifying the existing bridge for rail would be as costly as building an entire new three-mile bridge across the bay.

The net result of these errors—the bottom line—is that W.P. 2A.3 needs serious revision.


What can be the true explanation for these horrendous mistakes and inflated assumptions?

John Eddy of Ove Arup admitted that an error which fatally flaws the study was a mistake. Bonnie Nelson of Nelson/Nygaard stated that the structural assessment was a completed phase of the study, and that the numbers generated (i.e., more than three billion dollars) are the working assumptions for the rail study as a whole.

How can a fatally flawed assessment be retained as a working assumption? If the errors discussed above are honest mistakes, admitted since December 17, are there to be any corrections and adjustments?

An initial cost estimate was submitted to MTC by the consultants. (See Cost Estimates, attached.) The initial total was less than one billion dollars. We understand that preliminary figures can at times be too optimistic. However, a three- to four-fold increase over initial figures, which adds more than two billion dollars, seems rather drastic. Could everyone have been asleep, or could there possibly be a reason for this inflation?

For an answer, let us return to the Eastern Span Replacement Project, which first occasioned this opportunity to consider restoring rail service to the Bay Bridge.

The proposed eastern span (the design of which is not even finished) has been plagued for the last two years by charges of fraud and lack of safety.

Caltrans Bay Bridge Project Manager Dr. Brian Maroney stated publicly and to us personally that Amtrak-type train service on the Bay Bridge would ‘‘necessitate a different type selection’’ for the proposed eastern span. The project’s Draft Environmental Impact Statement (DEIS), required for federal review and published September 21, 1998, states that

BART or AMTRAK-type trains could not be accommodated on the East Span structure due to the combined length and weight of the trains.

Dr. Maroney’s statement and the DEIS give unequivocal testimony that the proposed eastern span is not capable of carrying heavy passenger trains, nor even lighter BART trains. Revelation that it would not safely accommodate trains—while the 66-year old western span could—would be an embarrassment to all involved: T.Y. Lin International and a consortium of architectural, engineering and construction industry consultants, MTC Commissioners and staff who have promoted the project over massive public opposition, Caltrans and MTC’s ‘‘blue-ribbon’’ panel of engineers and architects, the Engineering and Design Advisory Panel (EDAP).

MTC’s flawed study relied on input from a ‘‘Technical Advisory Committee’’ (peer-review panel) comprised of four proponents and designers of the proposed east-span with contractual conflicts of interest, five Caltrans employees with an agency investment in the east span, and only one consultant who has current experience in adding rail to a suspension bridge and who has no conflict of interest in the east span project. That one consultant, Kenneth Serzan of Parsons Transportation Group (see Chronology, attached) was not consulted. Mr. Serzan was included at the suggestion of Caltrans Bay Bridge Project Manager Brian Maroney because of Serzan’s involvement in designing the rail-deck retrofit of the Tagus River Suspension Bridge in Lisbon, Portugal, site of one of history’s most devastating earthquakes (1755). The construction contract for the Tagus bridge rail retrofit project, begun in 1996 and completed in August of this year, was $220 million. By comparison, $3.3 billion—the amount MTC publicized as the result of its rail feasibility study—would pay for numerous Tagus River rail retrofits—even with an adjustment for design and engineering, higher labor costs, and modifications to the Yerba Buena Island tunnel. The entire SFOBB, from Oakland to San Francisco, is roughly 2.6 times the length of the Tagus River bridge. Adjusting for length (2.6 x $220 million = $572), with liberal allowances for work on YBI and at the San Francisco landfall, the price for our longer, more complex SFOBB would be in the range of $800 million to $900 million. The original Ove Arup estimates ($758 million to $905 million) are confirmed by precedent.

What is the precedent of W.P. 2A.3?

Senator Quentin Kopp, author of the east span’s enabling legislation (S.B. 60), asked MTC (in a letter of August 10, 1998) for an estimate of cost associated with building a light or heavy rail line on the Bay Bridge, and an assessment of impacts to the east-span project. MTC’s Deputy Director William Hein dodged the question (Hein to Kopp letter, August 17, 1998) by stating it would cost $3 billion to rebuild an outdated 66-mile transit system, exclusive of all costs associated with the bridge itself. Other impacts cited by Hein (speaking for MTC staff leaders) are: likely delay in the east-span project of up to three years, and the possible addition of two to six dollars to existing bridge tolls. In fact, the Hein-to-Kopp letter suggests that the structural and cost assumptions were already made as early as August 1998:

‘‘Although there is no historical precedent for heavy rail service on the bridge [sic—an untrue statement], the cost of instituting such service could exceed the $3 billion estimate for light rail [emphasis added] for the following reasons:
[ ¶ ] ‘‘As noted above, the eastern span replacement would have to be redesigned. Two traffic lanes in each direction in the tunnel and on the existing suspension span would be required for rail service, unless the tunnel and western span were reconfigured to their pre-1958 state with rail on the lower deck and autos on the top deck—which would restrict the auto capacity of the bridge even further.
[ ¶ ] Moreover, since the suspension span originally was designed to accommodate light rail service, it is likely that the western half of the bridge would require further strengthening to carry heavy rail loads.’’ [ . . . ]

Hein’s letter makes a strong case for opposing rail. Speaking for MTC, he says in essence that it would be too difficult, too expensive, and in the end no one will ride it. The letter dates from August, 1998, as the issue of rail on the bridge was beginning to be debated prior to the November ballot initiative. The above passage reveals that Hein’s estimate for an entire 66-mile passenger system, which he wrongly equates to light rail, becomes, later in the letter, the expected cost of putting rail on the bridge.

By diverse methodologies, MTC arrives at a figure of $3 billion.

MTC’s reply to Senator Kopp was written during the campaign to include rail on the bridge, which would lead to its passage in November in Oakland, San Francisco, Emeryville and Berkeley—all four cities where it appeared on the ballot. Kopp, a principal proponent of MTC’s east-span replacement plans, was also the principal voice opposed to rail on the bridge. Evidently intending to sink all hopes for rail, he confidently cited MTC’s $3 billion figure in a radio debate with Mayor Bukowski of Emeryville, the rail initiative’s principal proponent.

While rail advocates won the electoral contest, they have not gotten past MTC, which now carries forward Hein’s August, 1998 analysis in remarkable detail in W.P. 2A.3.

A year and a half later, $3 billion is again the widely publicized price for rail on the bridge. That estimate is contrasted with the east-span replacement project, an ‘‘urgent seismic safety project’’ which would cost $1.5 billion. It is not publicized that, even as a drawing-board concept, the selected bridge type cannot work with rail. A new $1.5 billion bridge which cannot accommodate rail is publicly unacceptable, and so is the maintenance of contracts for its design and engineering.

Which leads to a disturbing possible explanation for the mistakes and conclusions of MTC’s Structural Assessment of Rail on the Bay Bridge: serious consideration of rail would result in professional embarrassment and termination of lucrative contracts for designing a bridge which cannot accommodate rail, which suffers from many other political and engineering nightmares, and which even its proponents have come to despise.

In light of all of this, it seems likely that MTC and Caltrans used a committee of conflicted individuals to derail any serious consideration of rail on the Bay Bridge. According to John Eddy, Ove Arup’s proposed rail design was altered considerably following review by the Technical Advisory Committee. Documents show that the cost was also subsequently inflated to over $3 billion from under $1 billion. The press duly reported (again) the $3 billion figure, this time officially certified by MTC’s consultants. The Legislature, the Governor and the public may suppose that such an extravagant expenditure ought to be rejected, and the rail option eliminated—thus salvaging the controversial and allegedly dangerous design for the new eastern span.

Naturally this is conjecture and speculation, for it is conceivable that the errors in W.P. 2A.3 were inadvertently overlooked by Ove Arup, then by Nelson/Nygaard, then by every one of MTC’s and Caltrans’ Technical Advisory Committee members and then again by the first two parties prior to final submittal to MTC, whose staff also did not notice the errors and publicly distributed the document and based their statements to the press upon it. And it is conceivable that Ove Arup’s initial figures, before being plagued by errors and the Technical Advisory Committee’s adjustments, were simply off by over two billion dollars.

Responses by Ove Arup, Nelson/Nygaard, MTC and Caltrans, and any other parties involved, to the questions raised here, is demanded by modern standards of integrity and public accountability. MTC should provide, further, a publicly announced and documented correction of the obvious errors found in the study and correction of the conclusions drawn from them. Not doing so will leave the MTC open to charges of gross mismanagement, or attempting to deceive the public.

We will post replies in their entirety at the website

Daniel Coman
Rick Feher
Robert Freehling

Technical Advisory Committee

home Initial cost estimates;
inflated cost estimates
Comparable project:
Tagus River Suspension Bridge