Professional Documents
Culture Documents
Design for Ship Collision with Bridges using AASHTO and CSA S6 - Clarifications and Recommendations
Design for Ship Collision with Bridges using AASHTO and CSA S6 - Clarifications and Recommendations
DESIGN FOR SHIP COLLISION WITH BRIDGES USING AASHTO AND CSA S6:
CLARIFICATIONS AND RECOMMENDATIONS
ABSTRACT
The design of bridges against ship collision is carried out in accordance with AASHTO LRFD Bridge
Design Specifications for bridges in the United States and CSA S6 Canadian Highway Bridge Design Code for
bridges in Canada. Both codes adopt an identical design methodology, which include a probabilistic analysis
to determine the annual likelihood of bridge collapse due to ship collision and comparison with pre-established
acceptance criterion that varies with bridge operational classification. This probabilistic approach is one of the
methods described and examined in PIANC’s 2001 INCOM Report of Working Group 19 titled Ship Collisions
due to the Presence of Bridges. In Canada, the CSA S6 code constitutes only a condensed version of
AASHTO’s requirements for vessel collision design, referring the reader to the American code when
information is not available in the Canadian publication. AASHTO also publishes a complementary technical
resource titled Guide Specifications and Commentary for Vessel Collision Design of Highway Bridges which
provides additional clarification and working examples for practitioners. There are several instances, however,
where these publications include ambiguous guidelines and supporting information. Examples of these include
the theoretical definition of the vessel collision load, the provisions behind the application of the collision force,
the reference to water levels, the correction factor for bridge location in navigation channels, and the
correction factor for cross currents. This lack of clarity can result in inconsistent implementation of the
methodology and potential application of its coefficients in a manner that is not intended by the codes. The
forementioned publications were reviewed, assessed, and compared to identify areas of potential
contradiction, lack of clarity, or other differences attributable to regional practices such as use of customary
units or technical nomenclature. This paper identifies key areas of potential improvement in the current
guidelines and provides recommendations for alternative language or additional clarifications where
appropriate.
1 INTRODUCTION
There are many bridge-design standards used around the world to design crossings that span navigable
waterways against potential collapse from vessel collisions with bridge components. Different methodologies
and provisions are available in national and international codes, guidelines, and regulations, which vary in the
probabilistic design requirements, estimation of impact load magnitudes, and load distribution requirements,
amongst other factors. Countries that have published codes or provide guidelines with specific provisions for
ship collision design include the United States, Canada, Australia, Sweden, Denmark, Norway, Germany, and
France. In the last decades, common framework provisions have been introduced in Europe via the Eurocode
(The State of Queensland, 2021; Pedersen, 2020; Gucma, 2009; Svensson, 2009; PIANC, 2001; Shaw,
2000). This paper evaluates and discusses the provisions used in North America to design bridges against
ship collision, with particular consideration of navigable waterway design practice.
analysis procedure (Method III). In general, Method II is expected to be used for all bridges except
when otherwise approved by the bridge owner or the authority having jurisdiction (AHJ);
• The design provisions are applicable for “normal merchant vessels, either steel hulled ship or barge
vessels” (AASHTO, 2009) with deadweight tonnage (DWT) not less than 1,000. Special-purpose
vessels, wood, or fiberglass constructed vessels, naval vessels, or recreational vessels are not
covered by the Codes;
• The acceptable annual frequency of bridge collapse, also referred to as the acceptance criteria, is
based on the bridge operational classification, with more stringent criteria applied to Critical or
Essential bridges than to Typical bridges.
Where C is a dimensionless empirically derived coefficient that varies depending on the system of units
used in the calculation. It’s worth noting that, despite it being expressed in units of mass (i.e. tonnes), the
DWT term is meant to represent a measure or characterization of the ship’s size, not its mass, nor its cargo
capacity or displacement tonnage. Detailed considerations on this topic, including evaluation of vessel mass
effects on the collision load, can be found in Pedersen (2020). The DWT corresponds to that of the design
vessel, which is obtained by one of the three available methods. For Method II, which involves the calculation
of the annual frequency of collapse, AF, the AASHTO Guide provides several examples for the selection of
the design vessel.
Risk Assessment Example 2.0 - Vulnerability Assessment of New Bridges states that “the equation for AF
is […] back-solved for each pier to determine the theoretical ultimate pier strength required to meet that pier’s
acceptance criteria. The design vessel for that particular pier then becomes the vessel in the fleet category
which is the closest match to the theoretical impact forces”. This statement implies that the ship collision force,
Ps, is calculated using the design vessel determined via probabilistic analysis for each particular pier or
element, which is consistent with the definition and formulation of Ps provided in the Codes.
Risk Assessment Example 1.8.3 - Method II Risk Analysis Summary, however, reads: “the procedure
used to compute Method II design impact force for each pier [is] to back-calculate the required pier resistance
strength (H) needed to meet the pier acceptance criteria for bridge collapse […]. This value of “H” then
becomes the theoretical design impact force for the pier and its foundations”. This statement implies that the
ship collision force is equal to the minimum theoretical pier resistance required to satisfy the probabilistic
threshold, which is not an uncommon way for practitioners to describe the collision force. However, this
approach implies that it would be acceptable for the ship collision force, Ps, to have a magnitude that falls
between design vessel categories, whereas both AASHTO LRFD Cl. 3.14.4 and CSA S6 Cl. A3.3.4 state that
a design vessel (or design vessel category) shall be selected for each pier or span component. It is also clear
that the collision force formulation for Ps requires the identification of a design vessel with a particular DWT, in
addition to other various geometrical attributes of the vessel used for the impact assessment. Therefore,
adopting a “theoretical” design impact force that does not correspond to a specific design vessel or design
vessel category does not seem appropriate, regardless of this force being used for the design of new bridges
or for the assessment of existing bridges.
Given the contradicting statements presented in Risk Assessment Example 1.8.3, additional clarification
could be included in the Codes to emphasize that the vessel collision force is not the same as the theoretical
impact force corresponding to the pier strength required to meet the acceptance criteria. It is also
recommended that further clarification is included in Risk Assessment Example 2.0, as the design vessel
should not only be the “closest match” but the “vessel that provides the closest match, but also produces a
higher collision force, to the theoretical impact force” (i.e. the designer should avoid ‘rounding down’ to a lower
design vessel category, as a bridge designed in such manner would not meet the annual frequency of
collapse requirement).
MHW is a tidal reference level used by the National Oceanic and Atmospheric Administration (NOAA) in
the United States, and although this reference water level is recognized by the Government of Canada’s
Department of Fisheries and Oceans (DFO), it is not commonly used in Canadian practice. For this reason,
MHW values are not readily available in standard publications by DFO.
MHW is defined as “The average of all the high water heights observed over the National Tidal Datum
Epoch” (NOAA, 2022), and there is no equivalent standard reference water level in Canada. Compared to
standard tidal definitions in Canada, the MHW level would fall somewhere between Mean Water Level (MWL)
and Higher High Water Mean Tide (HHWMT), which is defined as the “average of all the higher high waters
from 19 years of predictions” (DFO, Government of Canada, 2022). From a practical point of view, and given
the intent of the code to use typical waterway conditions, practitioners in Canada could adopt HHWMT in most
instances without incurring significant penalties for the increased water elevation, therefore avoiding the need
to calculate the MHW from first principles. It would be reasonable then for CSA S6 to change the term MHW
to HHWMT to be consistent with data readily available and published by DFO.
It would also be useful for AASHTO LRFD and CSA S6 to provide additional guidance on the definition of
“typical waterway conditions” for locations where water levels are influenced by river discharge. For example,
providing guidance on the percent of time exceedance that constitutes “typical” would be generally applicable
for all types of waterways, including those under both tidal and discharge influence, and would alleviate the
potential for inconsistent application amongst practitioners.
Without any mention of eccentric or oblique collision considerations, the adequacy of the current Codes’
methods to address such impact conditions may come into question, especially since other international
standards require the application of the impact force with explicit consideration of broadside impacts, including
a frictional impact force component that acts simultaneously with the lateral impact loads to account for
eccentric collisions (CEN, 2006). It would be therefore useful for both practitioners and bridge owners if the
Codes’ commentary clarified that glancing and eccentric collisions are already accounted for in the methods
prescribed by AASHTO LRFD and CSA S6.
A more commonly adopted interpretation of the definition provided by the AASHTO Guide is that there is a
single performance level for Critical bridges: functional performance after collision with the design vessel
(which happens to have a lower probability of occurrence than that considered for a Typical bridge). Note, for
example, that CSA S6 defines a Critical or Essential bridge simply as one that “must continue to function after
a vessel collision”. Similar definitions, stating that a Critical bridge must continue to function after a collision,
are shared by other relevant publications (Manuel, Kallivokas, Williamson, Bomba, & Cryer, 2006). However,
since the load factor, load modifiers, load combination, and load application requirements of the vessel
collision load are the same for Critical/Essential bridges as they are for Typical bridges, the interpretation that
the bridge “must continue to function after impact” from its corresponding design vessel would therefore have
to apply to all bridge classifications, and not only to Critical bridges. This interpretation would then contradict
the notion that the design objective of the code for extreme event limit states is “non-collapse of the structure”
(AASHTO, 2017), and that the design procedure for vessel collision, including the design vessel selection
methods, produces a design whereby piers, substructures, and connections to the superstructure are
“proportioned to prevent collapse of the superstructure” and where “damage or local collapse of substructure
and superstructure elements is permitted to occur provided that” redundancy or multi-load paths “exist in the
ultimate limit state to safely prevent superstructure collapse” so that the “structure element can be visually
inspected and repaired” (AASHTO, 2009).
A clear and definitive interpretation of the Critical bridge definition would be of great use to owners and
practitioners, especially if it offered a quantifiable description of its implied performance.
𝜃𝜃/90° is applied. The Codes provide no additional language to guide in the interpretation of the geometrical
arrangements shown in its figures related to bridge location. A typical representation of straight, bend, and
transition regions is shown in Figure 3.
Figure 2. Bridge Location Factor Schematic for Bend Regions as Presented in the Codes
There is generally no dispute on what factor should be applied when the bridge is located on a ‘straight’
region, sufficiently far away, at least 1 km (3,000 ft) from a bend region boundary, or when the bridge is
located on a ‘bend’ or ‘turn’ region. There are cases, however, when crossings are located between closely
spaced bend regions, where it may not be clear whether the bridge location should be considered on a ‘bend’
region or on a ‘transition’ region. One possible interpretation of the Codes’ figures is that a ‘transition’ region
may only exist between a ‘straight’ region and a ‘bend’ or ‘turn’ region, and if a ‘transition’ region were to exist
between ‘bend’ regions, then it should be at least 1 km (3,000 ft) long.
An extended definition that would be consistent with the design of navigation channels could state that: a
region should not be considered ‘straight’ unless a straight navigation path exists for at least 1 km beyond or
ahead of the ‘bend’ or ‘turn’ region; and more specifically, waterway zones with straight path navigation
between ‘bend’ regions, shorter than 5 x LOA (Length Overall) of the design vessel, should be treated as an
extension of the ‘bend’ or ‘turn’ region. This definition aligns well with navigation channel design practice, in
which “the designer should strive for an alignment consisting of a series of straight sections connected by
smooth bends” and that “if possible, the distance between successive bends should be greater than five ship
lengths” (PIANC, 2014). The 5 x LOA requirement may be relaxed to 3 x LOA if the consecutive bends turn in
the same direction. A graphical example of this best practice is provided in Figure 3.3 of PIANC (2014),
reproduced schematically in Figure 4 below.
In other words, navigation channel design practice aims to provide straight regions of navigation between
smooth bends, which implies that, as a general rule, ‘transition’ regions may be assumed to exist immediately
outside of all bend region boundaries provided adequate separation between bends exists in the waterway.
Notwithstanding, care should be exercised in areas where the design and improvement of navigational
channels have not kept pace with changes in the nature and characteristics of the world merchant fleet as
advised by the National Research Council (1993).
be considered, as sometimes, and specially as merchant vessels commonly have LOA’s averaging 200m, the
currents that influence the collision risk may exist several hundred meters away from the actual bridge
location.
PIANC (2014) states that “current speeds and directions can vary along the length of a channel,
especially at curves and channel intersections, and also with time. Special care is required to ensure
adequate channel width is provided where ships navigate at slow speed under cross-current". It also states
that "in order to keep a straight course under cross-currents, the ship should be operated to run obliquely to
the current, with the rudder amidships, to compensate for the current velocity perpendicular to the ship’s
desired course (i.e. the line of the channel)”. Similar ship navigation considerations are described in The
Japan Ship Owners' Mutual Protection & Indemnity Association (2014) when assessing navigation under
cross-wind influence for passage under bridges, for which a set leeway angle to starboard is required to
ensure safe passage, as depicted schematically in Figure 5.
Figure 4. Example of Leeway Angle to Starboard to ensure Safe Passage under Cross-Winds
Additional guidance on cross-current effects could be added to the Codes to incentivize a broader
assessment of the current vector components. New language may clarify that cross-current effects should be
included when the currents in the waterway have a tendency to veer the vessel off-course, altering its
intended trajectory along the navigation path on approach to the bridge, and especially under circumstances
that necessitate oblique navigation, with the rudder amidships, to compensate for the current velocity
component perpendicular to the ship’s desired course. Language of this sort would also address situations in
which the navigation channel follows natural flow features of the waterway (i.e. where prevailing currents are
naturally aligned with the channel), and where little correction is needed to keep the intended course, but
where localized current vector components may indicate otherwise, unnecessarily punishing the design.
Similar considerations may be applicable to cross-winds effects, although given the relative uniformity of wind
directions across relevant navigation distances on approach and under crossings, there appears to be little
need to provide further clarification in this regard.
It is a reasonable expectation that a protection factor would have a higher value when more protection is
provided. For example, a bridge pier with a number of protection dolphins will have a high level of protection
and one would expect, intuitively, to obtain a higher protection factor than that assigned to an unprotected
pier. This is not the case, however, as the Codes require the designer to first calculate the percent of
protection and then subtract this value from 1, resulting counterintuitively in a lower PF factor for cases where
more protection is given. This approach is required so that fully protected piers have zero annual frequency of
collapse (i.e. protection factor equals zero), and so that unprotected piers’ annual frequency of collapse
remains unaltered by the protection factor (i.e. protection factor equals to 1). This counterintuitive approach
could be easily remedied by adopting a different naming convention and updating the factor formulation.
The current methodology for calculating the level of protection of a given element utilizes a normal
distribution curve centered on the pier or element, with the percent of protection provided being equal to the
area beneath the curve bound by the vessel impact approach angles where protection is no longer achieved,
as shown in Figure 6. It stands to reason that this approach could continue to be adopted, but calculating the
area of exposure instead of protection, hence eliminating the need to subtract from unity in the PF equation.
PIANC America 2023
April 24-27, 2023, Fort Lauderdale, Florida, USA
The factor could then be renamed to the exposure factor, EF, in keeping with the intuitive argument: more
exposure, higher factor, and vice versa. We note that Knott (2018) also proposes this approach of quantifying
the exposure of a given pier rather than its protection. A revised formulae for an exposure factor is presented
in Equation 3.
% 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸
𝐸𝐸𝐸𝐸 = ( ) [3]
100
With this approach, fully protected piers would have an exposure factor of zero, resulting in a zero annual
frequency of collapse. The annual frequency of collapse would remain unaltered for a fully exposed pier, since
the exposure factor would be equal to 1. This simplified and straight forward approach can be achieved with
relatively minor updates to the Codes’ language and formulae.
1 𝑊𝑊
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 = ∗ ∗ 𝑉𝑉𝑉𝑉2 [4]
2 𝑔𝑔
It is therefore more reasonable for AASHTO LRFD to include some mention of this conversion, although
as stated (i.e. that the vessel collision energy equation is “the standard mV2/2 relationship for computing
kinetic energy with conversion from mass to weight”) is technically incorrect as the equation provided in the
code already considers the vessel displacement input in units of mass. Almost identical statements regarding
the “standard” kinetic energy expression can be found in published literature without the improper reference
for conversion of mass to weight, as stated by Manuel, Kallivokas, Williamson, Bomba, & Cryer (2006) when
describing AASHTO’s methods.
It is suggested to remove the language in the Codes that states “with conversion of mass to weight” as
this can lead to confusion and potential large discrepancy in the calculation of the vessel’s kinetic energy.
4 CONCLUSIONS
Specific provisions of AASHTO LRFD and CSA S6 were reviewed, assessed, and compared to identify
areas of potential contradiction, lack of clarity, or differences attributable to regional practices such as use of
customary units or technical nomenclature. Where appropriate, this paper provided recommendations for
possible improvement on how the information is currently presented in the Codes.
• The procedure for determining the magnitude of the collision force in method II was reviewed and
observations were made to the manner in which the AASHTO Guide describes the procedure and
defines the force magnitude;
• The definition of the water levels at which the collision force is to be applied were reviewed for
compliance with Canadian practice and recommendations were made for new conforming
nomenclature to be introduced in CSA S6;
• The purpose of the collision load application method, either parallel or normal to the direction of the
centerline of the navigable channel, was reviewed for its implied treatment of oblique and eccentric
collisions;
• Suggested language was provided for guidance on the treatment of cross-currents and correction
factors for bridge location used for calculating the probability of aberrance;
• A straight-forward formula for the protection factor treatment was proposed in accordance with
previously published literature.
In addition to these key aspects, other observations included improper definitions for the vessel
displacement tonnage and the vessel collision energy, the possible existence of implied performance of critical
bridges based on the definitions provided in the Codes, and the seemingly inconsistent classification
categories for bridges in Canada.
5 REFERENCES
AASHTO. (1991). Guide Specifications and Commentary for Vessel Collision Design of Highway Bridges.
Washington: AASHTO.
AASHTO. (2009). Guide Specifications and Commentary for Vessel Collision Design of Highway Bridges.
Washington: AASHTO.
AASHTO. (2017). AASHTO LRFD Bridge Design Specifications, 8th Edition. Washington: AASHTO.
CEN. (2006). EN 1991-1-7 (2006) Eurocode 1: Actions on Structures - Part 1-7: General Actions - Accidental
actions. Brussels: CEN.
CSA Group. (2019). CSA S6.1:19 Commentary on CSA S6:19, Canadian Highway Bridge Design Code. CSA
Group.
CSA Group. (2019). CSA S6:19 Canadian Highway Bridge Design Code. CSA Group.
DFO, Government of Canada. (2022, December). Tides, currents, and water levels. Retrieved from
https://tides.gc.ca/tides/en/definitions-content-tides-and-currents
Gucma, L. (2009). Methods of Ship-Bridge Collision Safety Evaluation, R&RATA # 2(13) part 1 (Vol. 2) 2009,
June . R&RATA.
International Code Council. (2020). 2019 California Building Code, Title 24, Part 2, Chapter 31F Marine Oil
Terminals. Washington: International Code Council.
Knott, W. (2018). Ship and Barge Collisions with Bridges over Navigable Waterways. Proceedings of the
PIANC-World Congress. Panama: PIANC.
Larssen, O. (1993). Ship Collision with Bridges - The Interaction between Vessel Traffic and Bridge
Structures. Zurich: IABSE - AIPC - IVBH.
Manuel, L., Kallivokas, L. F., Williamson, E. B., Bomba, M., & Cryer, a. (2006). FHWA/TX-07/0-4650-1 A
Probabilistic Analysis of the Frequency of Bridge Collapses due to Vessel Impact. Austin: Center for
Transportation Research at The University of Texas at Austin.
PIANC America 2023
April 24-27, 2023, Fort Lauderdale, Florida, USA
Military Surface Deployment and Distribution Command. (2013). STRAHNET ATLAS - Highways for National
Defense. Illinois: Transportation Engineering Agency.
National Research Council. (1983). Ship Collisions with Bridges - The Nature of the Accidents, Their
Prevention and Mitigation. Washington: National Academy Press.
NOAA. (2022, December). NOAA Tides & Currents. Retrieved from
https://tidesandcurrents.noaa.gov/datum_options.html
Pedersen, C. a. (2020). Design of Bridges Against Ship Collisions. Marine Structures - Design, Construction,
and Safety, Volume 74.
PIANC. (2001). Ship Collisions due to the Presence of Bridges - INCOM Report of WG 19. PIANC.
PIANC. (2014). Harbour Approach Channels - Design Guidelines. Report No 121 - 2014. PIANC.
The State of Queensland. (2021). Manual - Design Criteria for Bridges and Other Structures. Department of
Transport and Main Roads. The State of Queensland.
Shaw, O. a. (2000). Bridge Loads - An International Perspective. Spon Press.
Svensson, H. (2009). Protection of Bridge Piers against Ship Collision. Berlin: Ernst & Sohn.
The Japan Ship Owners' Mutual Protection & Indemnity Association. (2014). Preventing Damage to
Harbour Facilities and Ship Handling in Harbours - Vol. 32 Part 2. P&I.