Beruflich Dokumente
Kultur Dokumente
1.1 Foreword
The National Standards implementing Eurocodes will comprise the full text of
the Eurocode (including any annexes), as published by CEN, which may be
preceded by a National title page and National foreword, and may be followed by
a National annex. The National annex may only contain information on those
parameters which are left open in the Eurocode for national choice, known as
Nationally Determined Parameters, to be used for the design of buildings and
civil engineering works to be constructed in the country concerned, i.e.:
values and/or classes where alternatives are given in the Eurocode
values to be used where a symbol only is given in the Eurocode
country specific data (geographical, climatic, etc.), e.g. snow map
the procedure to be used where alternative procedures are given in the
Eurocode.
EXAMPLE 1-ATargetreliabilityindextest1
Solution: IntheLevelIIprocedures(seeFigureC1EN1990Overviewofreliabilitymethods),an
alternativemeasureofreliabilityisconventionallydefinedbythereliabilityindex
whichisrelatedtotheprobablyoffailure P f by:
Pf =
where isthecumulativedistributionfunctionofthestandardisedNormaldistribution.
ThegeneralformulafortheprobabilitydensityfunctionoftheNormaldistributionis:
x 2
exp ------------------ - x 2 2 2
2 2 e
f x = --------------------------------------- = ------------------------------
2 2
where isthelocationparameterand isthescaleparameter.Thecasewhere = 0 and
= 1 iscalledtheStandardNormaldistribution.Theequationforthestandardnormal
distributionis:
exp ---- x- 2 x-
----
2
2
2 e
f x = ------------------------ = ---------- .
2 2
Theprobabilityoffailure P f canbeexpressedthroughaperformancefunctiongsuch
thatastructureisconsideredtosurviveifg>0andtofailifg<0: P f = Prob g 0 .
IfgisNormallydistributed, istakenas = g g (where g isthemeanvalueofg,and g is
thestandarddeviation),sothat: g g = 0 and P f = Prob g 0 = Prob g g g .
Thecumulativedistributionfunction(CDF) ofarandomvariableistheprobabilityofits
valuefallingintheinterval ; ,asafunctionofx.
TheCDFofthestandardnormaldistribution,usuallydenotedwiththecapitalGreekletter ,is
theintegral:
4 7 x-2
----
1
e
2
= f x dx = ---------- dx .
2
END NOTE
For = 4 7 = 1 (1year):
4 7 x-2
----
1
e
2
4 7 = ---------- dx 10 6 .
2
Forareferenceperiodofn=100yearsthereliabilityindex n is(seeeq.C.3EN1990):
n = 1 n n = 4 7 100 10 6 100 = 10 4 n = 3 7 ,
1
where n = n istheinverseofthecumulativedistributionfunction.Thequantileof
thestandardnormaldistributionistheinverseofthecumulativedistributionfunction.
example-end
EXAMPLE 1-BApproachforcalibrationofdesignvalues(sectionC7EN1990)test2
Usingthegivennumericaldata,wefind(leadingvariableonly):
4
P E E d = + E = 3 90 10
5
P R R d = R = 6 15 10 .
Whentheactionmodelcontainsseveralbasicvariables,fortheaccompanyingactionsthe
designvalueisdefinedby:
P E E d = + E 0 4 ,
fromwhichweobtain:
2
P E E d = + E 0 4 = 0 28 = 0 28 4 8 = 1 344 = 8 95 10 .
example-end
EXAMPLE 1-CApproachforcalibrationofdesignvalues(sectionC7EN1990)test3
example-end
EXAMPLE 1-DApproachforcalibrationofdesignvalues(sectionC7EN1990)test4a
Solution: Consideringthesameassumptionsintheexampleabove(condition 0 16 E R 7 6
notsatisfied),itisseenthat:
E E E E E R R R R R
- 0,40 1,52 0,033 9,36 x 1,00 - 3,80 0,233 7,23 x
10 -1 10 -5
FromtableC3EN1990Designvaluesforvariousdistributionfunctions,byusingthe
Gumbeldistributionwiththegivennumericaldata,itfollowsthat:
0 577 0 577
a = -------------- = ---------- = 0 183 ; u = R --------------- = 30 --------------- = 26 85
R 6 7 6 a 0 183
= ---------
a = ------------- 577- = 30 0--------------
- = 1 283 ; u = 0-------------- 577- = 29 55 .
E
E 6 1 6 a 1 283
Therefore,itis(leadingvariableaction):
1 1 5 1
X di R = u --- ln ln R = u --------------- ln ln 7 235 10 = u --------------- 2 255
a 0 183 0 183
1
X di R = 26 85 --------------- 2 255 = 14 5
0 183
1 1 1 1
X di E = u --- ln ln E = u --------------- ln ln 9 357 10 = u + --------------- 2 712
a 1 283 1 283
1
X di R = 29 55 + --------------- 2 712 = 31 7 .
1 283
example-end
EXAMPLE 1-EApproachforcalibrationofdesignvalues(sectionC7EN1990)test4b
Given: Consideringthesameassumptionsintheexampleabove(seetab.1.1),derivethedesign
valuesofactioneffects E d withaprobabilityequalto 10 4 usingaNormalanda
Lognormaldistribution.
[Referencesheet:AnnexC][CellRange:A74:O74A140:O140].
example-end
EXAMPLE 1-F0factors(sectionC10EN1990)test5
Solution: ThedistributionfunctionsinTableC4refertothemaximawithinthereferenceperiodT.
Thesedistributionfunctionsaretotalfunctionswhichconsidertheprobabilitythatan
actionvalueiszeroduringcertainperiods.Thetheoryisbasedonthecalculationofthe
inversegammadistributionsprobabilitydensityfunctionoftheextremevalueofthe
accompanyingactioninthereferenceperiod.
Thegammadistribution,liketheLognormaldistribution,isatwoparameterfamilyofcontinuous
probabilitydistributions.Thegeneralformulafortheprobabilitydensityfunctionofthegamma
distributionis:
1
x - -
----------- exp x-----------
f x = ---------------------------------------------------------- ; x ; 0 .
where istheshapeparameter, isthelocationparameter, isthescaleparameter,and is
thegammafunctionwhichhastheformula:
t
1 t
= e dt .
0
Theinversegammadistributionsprobabilitydensityfunctionisdefinedoverthesupportx>0:
1 / x
g x ; 1 = ------------ x e .
1
Therefore,theinverse F x ofthecumulativedistributionfunction F x isthequantileofthe
1
standardgammadistribution: F x = x .
END NOTE
Ratioapproximatedtothenearestinteger: N 1 = T T 1 = 50 7 = 7 14 7 .
Shapeparameter (gammadistribution):
k = 2 = 1 V 2 = V 2 = 0 30 2 = 11 1 =
Scaleparameter (gammadistribution):
= -----------
= ----- - = V 2 = 0 30 2 = 11 1 = --1- = 11 1 1
2 V2 2
FromtableC4EN1990:
0 7 0 7 3 8 2 66
= 1
------------------------- = 1
------------------------------------- = 1
-------------------------- = 3 3
N1 N1 7
0 4 = 0 4 3 3 = 1 32 = 0 9066 ; 0 4 N1 = 0 9066 7 = 0 5034
0 7 = 0 7 3 8 = 2 66 = 0 9961 ; 0 7 N1 = 0 9961 7 = 0 9730 ;
ln 0 7 = ln 0 9961 = 0 0039 ,
0 4 = 0 4 3 3 = 1 32 = 0 0934
N 1 0 4 = 7 0 0934 = 0 6538
exp N 1 0 4 = exp 0 6538 = 0 5200
0 28 = 0 28 3 8 = 1 06 = 0 8563 ; ln 0 28 = ln 0 8563 = 0 1551
Quantiles(for = 11 1 , = 1 11 1 ):
1 N 1 1 N 1
F S 0 4 1 = F S 0 5034 = 0 9727 , F S 0 7 1 = F S 0 9730 = 1 6546
1 1
F S 0 7 = F S 0 9961 = 1 9797
1 1
F S exp N 1 0 4 = F S 0 5200 = 0 9850 .
SubstitutingthenumericaldatalistedaboveintoexpressionsintableC4EN1990,wefind:
a)Generalexpression:
1 N
F accompanying F S 0 4 1 9727- = 0 588 .
0 = -----------------------------
- = ------------------------------------------------ = 0-----------------
F leading 1
F S 0 7
N1 1 6546
b)Approximationforverylarge N 1 :
1
F accompanying F S exp N 1 0 4 0 9850
0 = ------------------------------ = --------------------------------------------------------------------- = ------------------ = 0 497 .
F leading 1
F S 0 7 1 9797
c)Normal(approximation):
F accompanying 1 + 0 28 0 7 ln N 1 V 1 + 0 28 3 8 0 7 ln 7 0 30 0 9106
0 = ------------------------------ = -------------------------------------------------------------
- = --------------------------------------------------------------------------------- = ------------------ =
F leading 1 + 0 7V 1 + 0 7 3 8 0 30 1 798
= 0 506 .
d)Gumbel(approximation):
F accompanying 1 0 78V 0 58 + ln ln 0 28 + ln N 1
0 = ------------------------------ = ------------------------------------------------------------------------------------------------------------------------
-=
F leading 1 0 78V 0 58 + ln ln 0 7
1 0 78 0 30 0 58 + ln 0 1551 + ln 7 1 0 78 0 30 0 58 1 8637 + 1 9459
= ------------------------------------------------------------------------------------------------------------ = --------------------------------------------------------------------------------------------------------- =
1 0 78 0 30 0 58 + ln 0 0039 1 0 78 0 30 0 58 5 5468
8450-
0-----------------
= = 0 391 .
2 1622
example-end
EXAMPLE 1-GD7.2Assessmentviathecharacteristicvaluetest6
Given: FindthedesignvalueofthepropertyXconsideringalreadyknowntheratio d m
betweenthedesignfactoroftheconversionfactorandthepartialfactorofthematerial.
Supposeasimplerandomsampleofsizen=30isdrawnfromapopulationhavingmean
andstandarddeviation (seetablebelow). Suppose the original distribution is
normal.
n xi n xi
1 19,3 16 17,3
2 19,8 17 19,2
3 20,1 18 22,4
4 20,4 19 16,0
5 20,3 20 15,0
6 19,3 21 15,6
7 18,0 22 18,2
8 17,4 23 17,4
9 21,3 24 19,2
10 19,4 25 16,3
11 20,2 26 15,3
12 20,5 27 14,0
13 21,0 28 13,0
14 22,3 29 15,3
15 18,5 30 16,5
Table 1.2 Sample results (n = 30). Reference Sheet: Annex D. Cell-Range B50:B64 - E50:E64.
Findthemean,variance,standarddeviationandthecoefficientofvariationofthe
samplingdistribution.Roundingtothefirstdecimal.
[Referencesheet:AnnexD][CellRange:A1:O1A82:O82].
Solution: Meanofthen=30sampleresults:
19 3 + 19 8 + 20 1 + 20 4 + + 13 0 + 15 3 + 16 5
m X = ------------------------------------------------------------------------------------------------------------------------------------------- = 18 3 .
30
Variance:
1
s x2 = --------------- 19 3 18 3 2 + 19 8 18 3 2 + 20 1 18 3 2 + + 16 5 18 3 2 = 6 0 .
30 1
Standarddeviation: s x = 6 0 = 2 45 .
Coefficientofvariation:
sx 45- = 0 13 .
V X = ------ = 2-----------
mx 18 3
Valuesof k n forthe5%characteristicvalueforn=30(seetab.D1RN1990):
1 67 V x known
kn =
1 73 V x unknown
DesignvalueofthepropertyX:
Xk n 14 3 V x known
- = -----d m x 1 k n V x = -----d 18 3 1 1 67 0 13 = -----d
X d = d -----------
m m m 1 73 m 14 2 V x unknown
havingconsideredalreadyknowntheratio d m .
example-end
EXAMPLE 1-HD7.2Assessmentviathecharacteristicvaluetest7
Given: Consideringthesamesampleresultintheexampleabove(seetab.1.2)andsupposing
the original distribution is Log-normal, find thedesignvalueofapropertyX
consideringalreadyknowntheratio d m .Roundingtothefirstdecimal.
[Referencesheet:AnnexD][CellRange:A84:O84A125:O125].
Estimatedvalue s for :
sy = s = ln V 2 + 1 V = 0 09 [input:(If V isknownfrompriorknowledge)].
Estimatedvalue s for [(If V isunknownfrompriorknowledge)]:
n
1 -
sy = s = -----------
n1 m
i y
2 =
i=1
1-
= ----- 2 960 2 897 2 + 2 986 2 897 2 + + 2 602 2 897 2 = 0 139 .
29
Valuesof k n forthe5%characteristicvalueforn=30(seetab.D1EN1990):
1 67 V x known
kn =
1 73 V x unknown
DesignvalueofthepropertyX:
15 6 V x known
X d = -----d exp m y k n s y = -----d exp 2 897 1 67 0 09 = -----d
m m 1 73 0 139 m 14 2 V x unknown
havingconsideredalreadyknowntheratio d m .
example-end
EXAMPLE 1-ID8.2Standardevaluationprocedure(Method(a))test8
Given: Developadesignmodelleadingtothederivationofaresistancefunction(seesection
D8.2EN1990Standardevaluationprocedure(Method(a))).Checkthevalidityofthismodel
bymeansofastatisticalinterpretationofallavailabletestdata(seetablebelow)(1).Verify
thatasufficientcorrelationisachievedbetweenthetheoreticalvaluesandthetestdata.
[Referencesheet:AnnexD][CellRange:A173:O173A384:O384].
r ti r ei r ti r ei
10,5 10,9 18,9 18,4
Table 1.3 Sample results (n = 30). Reference Sheet: Annex D. Cell-Range A194:B208 - D194:E208.
(1) rti theoretical values; rei experimental values from the tests.
Solution: Forthestandardevaluationprocedurethefollowingassumptionsaremade:
theresistancefunctionisafunctionofanumberofindependentvariablesX
asufficientnumberoftestresultsisavailable
allrelevantgeometricalandmaterialpropertiesaremeasured
thereisnostatisticalcorrelationbetweenthevariablesintheresistancefunction
allvariablesfolloweitheraNormaloralognormaldistribution.
Step1.Developadesignmodel,sayingeneral:
r ti = A i B i C I D I H I L I M I N I Q I T I .
Step2.Compareexperimentalandtheoreticalvalues.
Thepointsrepresentingpairsofcorrespondingvalues( r ti ; r ei )areplottedonadiagram
(seedataontable1.3):
Step3.Estimatethemeanvaluecorrectionfactorb.
r ei r ti = 10 5 10 9 + 12 6 12 3 + + 26 4 25 0 = 11401
i=1
r 2
ti = 10 5 2 + 12 6 2 + 14 7 2 + + 25 9 2 + 26 4 2 = 11501
i=1
r ei r ti
- = 11401
i=1
b = ------------------
n
--------------- = 0 991 .
11501
r ti2
i=1
Probabilisticmodeloftheresistance r = br t .Themeanvalueofthetheoreticalresistance
function,calculatedusingthemeanvaluesXmofthebasicvariables,canbeobtained
from:
r m = br t X m = bg rt X m .
Step4.Estimatethecoefficientofvariationoftheerrors.
Theerrorterm i foreachexperimentalvalue r ei shouldbedeterminedfromexpression
(D9EN1990):
r ei
i = -------
-.
br ti
Fromwhich,usingthegivennumericaldataintotable1.3,wefind(roundingtothree
decimalplaces):
r e1 10 9 - = 1 047 ; = ln = ln 1 047 = 0 046 ;
1 = --------
- = --------------------------------
br t1 0 991 10 5 1 1
r e2 12 3
2 = --------
- = --------------------------------- = 0 985 ; 2 = ln 2 = ln 0 985 = 0 015 ;
br t2 0 991 12 6
r e3 14 9
3 = --------
- = --------------------------------- = 1 023 ; 3 = ln 3 = ln 1 023 = 0 023 ;
br t3 0 991 14 7
...
r e30 25 0
30 = ----------
- = --------------------------------- = 0 956 ; 30 = ln 30 = ln 0 956 = 0 045 .
br t30 0 991 26 4
Substitutingtheabovenumericaldataintoexpressions(D.11),(D.12),(D13),wefind:
n n
1 1 0 046 0 015 + 0 023 + 0 045
= ---
n i = ---
n ln = i ----------------------------------------------------------------------------------------------- = 0 005
30
i=1 i=1
n
1 0 046 + 0 005 2 + 0 015 + 0 005 2 + + 0 045 + 0 005 2 = 0 001
2
s 2 = ------------ i = --------------------------------------------------------------------------------------------------------------------------------------------------------------------
n1 29
i=1
Step5.Analysecompatibility.
Thecompatibilityofthetestpopulationwiththeassumptionsmadeintheresistance
functionshouldbeanalysed.Ifthescatterofthe r ei ; r ti valuesistoohightogive
economicaldesignresistancefunctions,thisscattermaybereduced.Todeterminewhich
parametershavemostinfluenceonthescatter,thetestresultsmaybesplitintosubsets
withrespecttotheseparameters.Whendeterminingthefractilefactors k n (seestep7),the
k n valueforthesubsetsmaybedeterminedonthebasisofthetotalnumberofthetestsin
theoriginalseries.
Step6.Determinethecoefficientsofvariation V Xi ofthebasicvariables.
Consider,forexample,thedesignmodelforthetheoreticalresistance r ti asrepresented
bythefollowingrelation(bearingresistanceforbolts):
r ti = 2 5d i t i f ui 2 5 B i C i D i ;( A i = A = 2 5 = cos t ).
Theresistancefunctionabovecoversallrelevantbasicvariables X thataffectthe
resistanceattherelevantlimitstate.Thecoefficientsofvariation V Xi willnormallyneed
tobedeterminedonthebasisofsomepriorknowledge.Therefore,letussay:
1)coefficientofvariation V d = 0 04 ofthebasicvariableoftheboltsdiameter;
2)coefficientofvariation V t = 0 05 oftheb.v.ofthethicknessoftheconnectedpart;
3)coefficientofvariation V fu = 0 07 oftheb.v.oftheultimatetensilestrengthofthe
materials.
Step7.Determinethecharacteristicvalue r t oftheresistance.
Theresistancefunctionforj(=4)basicvariablesisaproductfunctionoftheform:
r = br t = b A B C D .
Coefficientofvariation V r :
j
V2r = V 2 + 1 V + 1 1 = V 2 + 1 V a2 + 1 V d2 + 1 V t2 + 1 V fu + 1 1
2 2
Xi
i=1
V = exp s 2 1 = 0 032
n
V rt2 = V = V a2 + V d2 + V t2 + V fu = 0 + 0 04 2 + 0 05 2 + 0 07 2 = 0 009 .
2 2
xi
i=1
Thenumberoftestislimited(n=30<100).Inthiscasethecharacteristicresistance r k
shouldbeobtainedfrom[seeequation(D.17)EN1990]:
r k = bg rt X m exp k rt Q rt k n Q 0 5Q 2 with:
Q = ln V r2 + 1 = ln 0 01 + 1 = 0 100
Q rt = ln V rt2 + 1 = ln 0 009 + 1 = 0 095 ; rt = Q rt Q = 0 095 0 100 = 0 95
Q = ln V 2 + 1 = ln 0 032 2 + 1 = 0 032 ; = Q Q = 0 032 0 100 = 0 32 .
Valuesof k n forthe5%characteristicvalueforn=30(seetab.D1EN1990):
k 1 64 for n
=
kn 1 73 V x unknown
Substitutingthenumericaldataintoexpressionsabove,wefindthecharacteristicvalueof
theresistance:
r k = r m exp k rt Q rt k n Q 0 5Q 2 =
= r m exp 1 64 0 95 0 095 1 73 0 32 0 032 0 5 0 100 2 = r m exp 0 171 =
= r m exp 0 171 = r m 0 84
Herethecharacteristicvalue r k isrepresentedasbeingproportionaltoitsmean r m .
example-end
EXAMPLE 1-JD8.3Standardevaluationprocedure(Method(b))test9
Given: Consideringthesameassumptionsintheexampleabove,determinethedesignvalueof
theresistancebytakingaccountofthedeviationsofallthevariables.
[Referencesheet:AnnexD][CellRange:A387:O387A413:O413].
Solution: InthiscasetheprocedureisthesameasinD8.2,exceptedthatstep7isadaptedby
replacingthecharacteristicfractilefactor k n bythedesignfractilefactor k d n equaltothe
product R assessedat 0 8 3 8 = 3 04 ascommonlyaccepted(seeAnnexCEN1990)
toobtainthedesignvalue r d oftheresistance.
Forthecaseofalimitednumberoftests(hereinn=30<100)thedesignvalue r d shouldbe
obtainedfrom:
r d = bg rt X m exp k d rt Q rt k d n Q 0 5Q 2
where:
k d n isthedesignfractilefactorfromtableD2forthecase V X unknown
k d isthevalueof k d n for n [ k d = 3 04 ].
Thevalueof k d n fortheULSdesignvalue(leading)is3,44(seetableD2EN1990).
Therefore,weget:
r d = r m exp k d rt Q rt k d n Q 0 5Q 2 =
= r m exp 3 04 0 95 0 095 3 44 0 32 0 032 0 5 0 100 2 = r m exp 0 315 =
= r m exp 0 315 = r m 0 73
havingrepresented r d asbeingproportionaltoitsmean.
Dividingthecharacteristicvaluebythedesignvalueweobtain:
r r m 0 84
R = ----k = --------------------- 1 15
rd r m 0 73
havingestimated V fromthetestsampleunderconsideration(seedataintab.1.3).
example-end
EXAMPLE 1-KD8.4Useadditionalpriorknowledgetest10
example-end