See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/346463107

How to draw an environmental contour using IFORM (Inverse First-Order

Reliability Method)

Method · November 2020

CITATIONS READS

0 151

1 author:

Hyeong Jin Kim

Pusan National University
2 PUBLICATIONS   0 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Hyeong Jin Kim on 28 December 2020.

The user has requested enhancement of the downloaded file.

 Environmental contour using IFORM (Inverse First-Order Reliability Method)

Hyeong Jin Kim

[email protected]

Pusan National University

This paper is for students, designers or researchers who are not familiar with IFORM.

IFORM, suggested by Winterstein (Winterstein et al. 1993) for reducing structural reliability
problem, is a useful method to determine an extreme sea state condition by using environmental
contours. The method uses a joint distribution function which represents dependently
significant wave height H S and zero-up-crossing wave period TZ defined by Equation (1).

f H S TZ ( hs , t z ) = f H S ( hs ) fTZ HS (t z hs ) (1)

A common practice for fitting joint distribution to the wave data, such as wave scatter diagram,
is to fit a marginal distribution to H S , and a conditional distribution to TZ (Velarde et al. 2019,
DNV 2014). Generally, Weibull distribution is used for the marginal distribution of H S ,
lognormal distribution is used for conditional distribution of TZ . Equation (2) – (4) represent
Weibull, 3-paramether Weibull, and lognormal distribution respectively.

 1 −1
 h    h  1 
f H (hs ) = 1  s  exp  −  s   (2)
S
1  1    1  

 2 −1
  h −2    h −   2 
f H (hs ) = 2  s  exp  −  s 2
  (3)
S
2  2     2  

1  −(ln(t z ) −  (hs )) 2 
fTZ (t z hs ) = exp   (4)
HZ
t z  (hs ) 2  2 2 

Where i , i ,  i ,  (h),  (h) are scale, shape, location, mean, and standard deviation parameters.
For the conditional distribution, parameters of each lognormal distribution corresponding
 interval of hs should be obtained (Vazirizade et al. 2019); in case of Appendix Table 1,
there are 28 rows for each lognormal distribution corresponding interval of hs . After all,
lognormal parameters,  (h) and  (h) , can be fitted by using Equation (5), (6) and some
estimation method such as MOM(Method of moment), MLE(Maximum likelihood
estimation) and LSM(Least square method). Figure 1 illustrates fitted curves of
 (h) and  (h) .

 (h) = a1 + a2 h a 3
(5)

 (h) = b1 + b2 exp(b3h) (6)

b1 b2 b3
0.051 0.181 - 0.181
Lognormal parameter,

Lognormal parameter,

a1 a2 a3
- 0.956 2.518 0.122

Significant wave height, Hs (m) Significant wave height, Hs (m)

Figure 1. Fitted curves of lognormal parameters

The main idea of IFORM, which is based on the structural reliability approach, is to know
probability of exceedance and to seek the corresponding response level (Haver and Winterstein,
2009). The failure probability corresponding a return period TR is defined by Equation (7).
Thus, the reliability is defined by Equation (8).

1 1
pF = q = (7)
 24   24 
365    365     TR
 ts   ts 

rR = 1 − pF (8)
Where pF is the failure probability, rR is the reliability, q is an annual exceedance
−2 −4
probability; NORSOK recommends q of 10 in ULS and q of 10 in ALS. t S is time
interval of recorded data. For example, when if the data is recorded every hour and the return
period is 100 years, the failure probability and the reliability is calculated as follows:

1
pF = = 1.14  10 −6 (9)
 24 
365    100
 1 

r100 = 1 − 1.14 10−6 = 0.9999989 (10)

 is corresponded to the reliability in the CDF(Cumulative distribution function) of the

standard Gaussian function as shown Figure 2. In the U-space, target events related to TR
must be on the circle with a radius  (Haver and Winterstein, 2009). The relationship among
reliability, reliability index  , and variables of U-space (or the standard Gaussian variable
space) is defined by Equation (11).

Standard Gaussian CDF

rR
u2

Unsafe
u2
Probability

u1
u1

pF

Figure 2. CDF of standard Gaussian distribution and U-pace

 −1 (rR ) =  = u1i 2 + u2i 2 (11)

Where  is the cumulative distribution function of the standard Gaussian distribution
function, ui is the variables of the U-space.

To estimate the environmental contour line, the transformation procedure is needed from the
U-space to the environmental parameter space consisted of H S and TZ coordinate. The
procedure is an inverse process of determining a radius  . The variables of U-space can be
transformed to environmental parameter space as shown in Figure 3 and Equation (12). The
inverse procedure of this transformation is called Rosenblatt transform (Rosenblatt, 1952).

Standard Gaussian CDF

Probability

U
(a)

Lognormal CDF 3-parameter Weibull CDF

Probability

Probability

(sec)
(b)

Figure 3. (a) Transformation from ui to ki , (b) Transformation from ki to ti and hi

ti = CDFTZ HS
−1
 ( u ) ,
1i hi = CDFH S −1  ( u2i ) (12)
For the conditional distribution, the physical value, generally TZ , should be transformed by
using each CDF of lognormal distribution corresponding hs as mentioned above. Equation
(13) – (15) is CDF of Weibull, 3-parameter Weibull, lognormal distribution function
respectively. Finally, connected points on the environmental parameter space is an
environmental contour line which has the constant exceedance probability as shown Figure 4
(Choi et al. 2019).

1

  hs   
FWeibull (hs ) = 1 − exp −    (13)

  1 
 
2

  hs −  2   
F3 p −Weibull (hs ) = 1 − exp −    (14)
2  
 
 

1  ln t −  
Flog normal (t z ) = erfc  − z  (15)
2   2 

u2
Significant wave height, Hs (m)

100-year contour

u1 u1

Average zero-up-crossing wave period, Tz (sec)

Figure 4. U-space and environmental contour

-END-
Nov. 2020.
References

1. Choi, Jang, Park, Kim, Park. 2019. Improved environmental contour method based on an
optimization of hybrid models. Applied Ocean Research 91: 101901.
2. DNV. 2014. Environmental conditions and Environmental Loads, DNV-RP-C205.
3. Eckert-Gallup, Martin. 2017. New methods for calculating environmental contours of
extream sea states. Sandia National Laboratories.
4. Haver, Winterstein. 2009. Environmental contour lines: a method for estimating long-term
extremes by a short-term analysis. Transactions of the Society of Naval Architects and Marine
Engineers 116: 116-127.
5. NORSOK. 2007. Actions and Actions Effects, NORSOK N-003.
6. Rosenblatt. 1952. Remarks on a multivariate transformation. Ann Math Statist 23: 470-472.
7. Vazirizade, Haldar, Gaxiola-Camacho. 2019. Uncertainty quantification of sea waves – an
improved approach. Oceanography & Fisheries 9.

8. Velarde, Vanem, Karmhøft, Sørensen. 2019. Probabilistic analysis of offshore wind turbines
under extream resonant response: application of environmental contour method. Applied
Ocean Research 93: 101947.
9. Winterstein, Ude, Cornell, Bjerager, Haver. 1993. Environmental parameters for extreme
response: inverse FORM with omission factors, Proceedings of the ICOSSAR 93.
 Appendix

[Scatter table]
Hs/Tz 0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 Total
13.75 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10
13.25 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10
12.75 0 0 0 0 0 0 0 0 0 0 10 0 0 0 10
12.25 0 0 0 0 0 0 0 0 0 0 10 0 10 0 30
11.75 0 0 0 0 0 0 0 0 0 0 10 10 10 0 30
11.25 0 0 0 0 0 0 0 0 0 0 20 20 10 0 40
10.75 0 0 0 0 0 0 0 0 0 10 20 20 10 0 60
10.25 0 0 0 0 0 0 0 0 0 10 30 30 10 0 80
9.75 0 0 0 0 0 0 0 0 0 30 50 30 10 0 120
9.25 0 0 0 0 0 0 0 0 0 60 80 30 10 0 180
8.75 0 0 0 0 0 0 0 0 10 100 120 30 10 0 270
8.25 0 0 0 0 0 0 0 0 30 160 110 20 10 0 330
7.75 0 0 0 0 0 0 0 0 90 260 130 30 0 0 510
7.25 0 0 0 0 0 0 0 10 200 390 130 40 10 10 790
6.75 0 0 0 0 0 0 0 20 410 450 150 40 10 0 1100
6.25 0 0 0 0 0 0 0 100 690 450 160 50 10 10 1470
5.75 0 0 0 0 0 0 0 350 990 410 180 60 20 10 2030
5.25 0 0 0 0 0 0 20 890 1170 440 220 70 30 0 2850
4.75 0 0 0 0 0 0 170 1780 1230 520 250 90 10 0 4070
4.25 0 0 0 0 0 10 780 2420 1110 560 320 100 0 0 5300
3.75 0 0 0 0 0 60 2270 2450 1220 660 350 70 0 0 7090
3.25 0 0 0 0 10 580 3750 2190 1310 800 260 30 10 0 8920
2.75 0 0 0 0 30 2960 4070 2220 1590 730 130 10 0 0 11760
2.25 0 0 0 10 660 6040 3560 2300 1360 300 30 0 0 0 14260
1.75 0 0 0 40 4810 6240 3430 2010 470 40 0 0 0 0 17030
1.25 0 0 10 1550 7890 4190 2380 560 20 0 0 0 0 0 16590
0.75 0 0 70 2410 1480 940 130 0 0 0 0 0 0 0 5020
0.25 0 0 0 0 20 0 0 0 0 0 0 0 0 0 20
Total 0 0 80 4020 14900 21030 20550 17290 11900 6380 2790 800 200 50 1×105

Table 1. Example of wave scatter diagram

View publication stats