Changing Practice in Gulf of
Mexico Design and Operating
Criteria
The Early Days
(sometimes it is better to be lucky than good)


Gulf of Mexico platforms built around 1946 – 48 typically
had deck heights 20 to 40 feet above mean sea level
Consultants’ evaluations:




“in 100 feet of water waves will probably seldom, IF EVER, exceed
20 feet in height”
“ … settled on a maximum wave height of about 25 feet and a
recommended deck height of 32 feet”
Consensus: maximum wave ~ 29 feet, occurring perhaps
once every 40 to 50 years.
Very loose “consensus” – no API guidance, little regulation.
Luck Starts to Run Out

1947 – 1952: series of relatively weak, small
hurricanes in the Gulf

October 1949 – platform off Freeport damaged –
post-mortem suggested waves as high as 40 feet.


Observed damage in others led to estimates of
22-29 feet – calls into question both the upper
limit and frequency of occurrence of high waves
in Gulf
Leads to stronger designs for a few operators
Lulled to Sleep?
All Hurricanes and Trop Storms in Gulf 1947 - 1955
Luck Starts to Run Out

1956 – Hurricane Flossie

A Category 1 storm


50 men rode out the storm in the Gulf. One
vessel lost its anchor and floated around during
the storm in keeping with a philosophy of “taking
a calculated risk that they would be safe.”
Led to calls for complete evacuation in
hurricanes.
Luck Starts to Run Out

1957 – Hurricane Audrey

Forms in Gulf – now called “Sudden storms”


One mobile drilling rig sank, with four tenders
suffering damage when pulled loose from their
mooring and running aground
Industry record of no fatalities held.
Industry Intensifies Action




1957 – Hurricanes Audrey and Bertha – three
significant storms in 2 years.
API forms “Advisory Committee on Fundamental
Research on Weather Forecasting.”
Disbanded in 1962.
Why? Other issues and Gulf fairly quiet (Carla in
1961 but it hits Texas)
Hmmmm?
All Hurricanes and Trop Storms in Gulf 1956 - 1963
No Consensus

Deck height practices:



Varied from the 1950 era standard of 28 –
32 feet above mean Gulf level to higher
than 50 feet.
Not coincidently, those using higher values
were companies directly impacted by storms
either in terms of property or direct threat
to employees.
Higher meant safer and more expensive –
“each company placed a bet on the right
combination of safety and cost”.
(Primitive cost-benefit analysis)
Luck Starts to Run Out
All Hurricanes and Trop Storms in Gulf 1964 - 1973
Luck Run Outs 1964 - 1969

1964: Hurricane Hilda – Category 4
http://en.wikipedia.org/wiki/Hurricane_Hilda

1965: Hurricane Betsy
http://en.wikipedia.org/wiki/Hurricane_Betsy
Luck Runs Out 1964 - 1969


1964: Hurricane Hilda – Category 4
Hilda was the most damaging tropical cyclone to
the offshore oil industry, at the time of its impact.

More than US$100 million in losses.

13 oil platforms were destroyed

5 more damaged beyond repair[
Luck Runs Out 1964 - 1969



1965: Hurricane Betsy – strong Category
3 at landfall
Eight offshore oil platforms were
destroyed during Betsy, with others
experiencing damage.
The oil rig Maverick disappeared during
the cyclone
Industry Action

1966: API Committee on Standardization
of Offshore Structures created.


Focus to create better design standards
through cooperative efforts.
Basic research and measurement of wind,
waves, and soils continues.

Includes the Ocean Data Gathering Program
(ODGP) – 6 platforms instrumented in Gulf
from 1968 through 1971
Some “Good Luck”, Some Bad

1969: Hurricane Camille
http://en.wikipedia.org/wiki/HurricaneCamille
ODGP measured a wave
between 70 and 75 feet high!!!
Some “Good Luck”, Some Bad

1969: Hurricane Camille
Used to calibrate hindcast
models in Gulf for decades.
Metocean criteria developed
using those hindcasts as
database of storms in Gulf.
ODGP measured a wave
between 70 and 75 feet high!!!
API Standards and Design Waves


First API offshore standard (RP2A) issued
in 1969
No design wave information until 7th
edition in 1976.
• Recommends use of “the 100-year
wave”

To this point the owner chose the return
period and the use of both 25 and 100 year
values was common.
• 1% risk of exceedance annually = 100 yr
• 4% risk of exceedance annually = 25 yr
API Standards and Design Waves


20th edition (1993) includes a new wave
force calculation “recipe” that substantially
changes that of the 19th edition (1991)
Design (“100-year”) wave conditions
changed as well
WOW!

1992 saw Hurricane Andrew
http://en.wikipedia.org/wiki/Hurricane_Andrew
What Hath Andrew Wrought?






Category 4 storm in Gulf
MMS estimates 700 structures took
“significant” hit
22 older platforms destroyed
65 others with significant damage
Majority had been designed to 25 year
values and 35 to 40 foot decks
Newer platforms that were designed with
decks to pass up to 72 foot waves had
only minor damage
Effect of Andrew and New API
Force Recipe


Gave a boost of energy to an API
committee looking at assessment of
existing platforms
Task group decided that the new criteria
should be relaxed for existing platforms
and consideration be given to
CONSEQUENCE OF FAILURE

Willing to take higher risks with older assets in part
because cost to modify/replace are too high
Effect of Andrew and New API
Force Recipe

Three categories with different metocean criteria:

L-1 (high consequence / manned-evacuated or unmanned
• Full-population of hurricanes, 1% annual exceedance probability

L-2 (low consequence / manned-evacuated)
• Sudden hurricane and winter storm population

L-3 (low sequence / unmanned, or minimum consequence
• Winter storm population

Included deck height criteria
Consequence-based Criteria for
New-Build Platforms


CBC for assessing existing platforms in
place in 1996 (issued as a supplement to
the 20th edition of RP2A.
Sets stage for introducing this concept for
new-build



L1
L2
L3
Table 1 – Risks Considered for
Consequence-Based Criteria for the
Gulf of Mexico
CONSEQUENCES
OF FAILURE
LEVEL 1
LIFE SAFETY
1
Manned, nonevacuated
HIGH
2
Manned,
evacuated
MEDIUM
3
Unmanned
LOW
Lulled to Sleep Again?
All Hurricanes and Trop Storms in Gulf 1993 - 2000
What’s Happening Out There?
All Hurricanes and Trop Storms in Gulf 2001 - 2005
Impact of Ivan, Katrina, and Rita
Impact of Ivan, Katrina, and Rita



2005 Atlantic Season: most active in
recorded history
28 named storms, 15 hurricanes, 7 major
hurricanes, and four category 5 hurricanes
(per NOAA NHC)
Worst season previously: 1933 with 21
named storms, 10 hurricanes, and 5
major hurricanes
Impact of Ivan, Katrina, and Rita



2004-2005 was worst two year period (23
hurricanes) since 1886-1887 (21 total)
Lots of damage to platforms and mobile
rigs (114 platforms destroyed in Katrina
and Rita)
Lots of infrastructure damage (pipelines)
leading to loss of oil and gas production
MODU Failures – Floating (Semi) and Jackup
Ivan, Katrina, and Rita
Total Failure Partial Failure
Ivan: Sept. '04, Cat 4
Katrina: Aug. '05, Cat 5
Rita: Sept. '05, Cat 4
Semi
4
1
Jackup
1
1
Semi
4
3
Jackup
2
0
Semi
7
4
Jackup
7
1
API Metocean Reaction



Study of Ivan led to conclusion it was a
rare event statistically but no need for
significant criteria revision (OTC Paper
17740)
Recommendation was to simply include
Ivan in any extremal analysis
Maximum Hs (m):



Ivan
Katrina
Rita
16.0
16.9
11.5
API Criteria before these storms?
Hmax = 21.5 m
Hs
≈ 12.6 m
Industry and API Metocean Reaction


The rapid-fire occurrence of three huge storms in
two years led to significant revisions in part due
to a mooring risk JIP led by ABS which required
the best possible metocean data
Several key findings:
• Loop current and or Loop eddies provided get source of
energy for all three of these storms
• Dividing the Gulf into four regions was deemed
appropriate
• Use of the full 1990 – present hurricane database was
not appropriate
Depth below
Ocean surface, m
Loop Current - Source of Deep Warm Water

0
-100
-200
-300
-400
Other
Areas
Loop
and
Eddy
Areas
10° 20° 30° 40°
Temperature, deg C
Stages develop on time scale of months = persistent warm water conditions
Storm Tracks over Loop
http://www.esl.lsu.edu/qu
icklinks/publications/imag
es/PreStorms.gif
Reprinted courtesy of
Colorado Center for
Aerodynamics
Research
API Metocean Criteria Changes
after Ivan, Katrina, and Rita

Gulf divided into 4 regions.
Bias in Storms Prior to 1950


Measurements were sparse, often only at
land stations. What went on in the Gulf
was speculative.
In a 2006 paper (OTC 18418), Cooper and
Stear concluded that there was a negative
bias in the 1900 to 1949 storms as
characterized by the National Hurricane
Center.
Bias in Storms Prior to 1950
Start=1900, End=1949, Min CPI (mb)=63
200
400
40
60
80
100
CPI (mb)
120
140
Start=1955, End=2005, Min CPI (mb)=63
0
1900/09/06
1915/09/30
1916/08/19
1919/09/12
1932/08/14
1933/09/04
Distance from Shore (km)
Distance from Shore (km)
0
160
1961/09/10
1964/10/02
1965/09/09
1969/08/17
1970/08/02
1974/09/09
1979/09/12
1980/08/10
1982/09/10
1992/08/25
1995/10/10
1999/08/22
2004/09/15
2005/07/09
2005/08/27
2005/09/22
200
400
40
60
80
100
CPI (mb)
120
140
160
Plots provided by Chevron Energy Technology Company
Plots:
6 pre-’50; 16 post-’50 storms
Pre-’50, 70% of storms show no drop as they near coast
Post-’55, 12% show no drop
Suggest: pre-’50 applied coastal data offshore resulting in low bias
Other factors support conclusion:
NOAA generally does not use pre-’50
8 of top 10 winds occurred post-’50
7 of top 10 waves occurred post-’50
Hurricane Reconnaissance
Became common after WWII
Industry Response
API RPs 2I, 2SM, and 2SK were
all updated
MODU Mooring JIP (budget
approx. $2.2m)
Interim Guidance for Design of Offshore Structures for Hurricane Conditions
API BULLETIN 2INT–DG
First Edition, May 2007
Interim Guidance for Assessment of Existing Offshore Structures for Hurricane
Conditions
API BULLETIN 2INT–EX
First Edition, May 2007
Interim Guidance on Hurricane Conditions in the Gulf of Mexico
API BULLETIN 2INT–MET
First Edition, May 2007
Measured Data is Foundational

Hindcasts are used to generate
criteria BUT these are calibrated
against data
Excellent network of data buoy in Gulf and along coast
Future Hurricanes




More/less?
Stronger?
Similar tracks?
Was the 2004-2005 season a
precursor of things to come?
Future Hurricanes

Atlantic Hurricane Variability Over Time: 1886 – 2004


Multi-decadal variability
El Nino
Yellow = named storms, Green = hurricanes, Red = Category 3 and above
Future Hurricanes


Was the 2004-2005 season a precursor of things
to come? Was it all that unusual?
How confident can we be in the historical record?
• Prior to about 1950 – no air reconnaissance
• Satellites much later
Future Hurricanes


Key factors: warm water and wind shear
generally considered most important
Area of much current research AND
considerable controversy

“Experts” do not agree

Models do not agree
Two interpretations of SST data lead to VASTLY
different future Atlantic activity
extrapolated into the 21st century
using absolute SSTs calculated from
global climate model projections
suggest that it is the SST in the
tropical Atlantic main
development region relative to the
tropical mean SST that controls
fluctuations in Atlantic hurricane
activity
Science 31 October 2008:
Vol. 322. no. 5902, pp. 687 689
DOI: 10.1126/science.1164396
CLIMATE CHANGE:
Whither Hurricane Activity?
Gabriel A. Vecchi,1 Kyle L.
Swanson,2 Brian J. Soden3
Future Hurricanes and API



As opposed to the general historical pattern of
API, being REACTIVE, there is a shift to being
more PROACTIVE
Funding a synthetic hurricane study at a cost
significantly larger than “normal” API-funded
research to develop a long term (100000 yrs)
synthetic data base of hurricanes in the Gulf
Supportive of National Center for Atmospheric
Research (NCAR) modelling work on how climate
change is likely to influence hurricane activity
through 2055 through RPSEA with Industry reps
on steering committee, members’ time, etc.
Future Hurricanes and API


No criteria changes to account for any
future climate scenario being applied in
part due to uncertainties
May be applied when the results of the
sponsored research efforts are finished
QUESTIONS ?
BACKUP
TABLE 2 – LIFE SAFETY
LEVEL
MANNING
CONDITIONS
1
Manned, nonevacuated
2
Manned,
evacuated
3
Unmanned
EXPOSURE TO PERSONNEL
Platform continuously manned
 Evacuation prior to design event is not
intended or is
not feasible
 Personnel exposed to severe hurricanes

Platform normally manned
 Evacuation prior to design event is planned and feasible
 Largest events that personnel are exposed to are a
“sudden” hurricane or a winter storm

Platform normally unmanned except for day trips to
perform short field operations
 May have emergency shelter but no permanent quarters
 Personnel exposed to worst event that can occur during
the day




Occasional manning for short durations
Short maintenance, construction, workover, or drilling
Scheduled to minimize potential hurricane exposure
TABLE 3 – CONSEQUENCE OF FAILURE
PLATFORM CHARACTERISTICS
PLATFORM
CATEGORY
CONSEQUENCE
OF FAILURE
1
High

2
Medium

3
Low

Platform with major drilling, production, pipeline
processing, or storage facilities
 Other platforms with high consequences of failure
 Wells not shut-in during design event
 Potential for well flow in the event of platform failure
 All platforms in depths greater than 400 feet
Platform with drilling or production pipeline facilities
with medium consequences of failure
 Wells shut-in during design event
 Wells protected by SSSV’s
 Oil storage limited to process inventory and “surge”
tanks for pipeline transfers
Small well protectors and caissons with small
consequences of failure
 No more than 5 completions on or connected to
platform
 Wells shut-in during design event
 Wells protected by SSSV’s
 No more than 2 pieces of production equipment
 Oil storage limited to process inventory
 Maximum water depths of 100 feet