Uncertainty and Expert Assessment for Supporting Evaluation of Levees Safety

Testo completo

(1)E3S Web of Conferences 7, 03019 (2016) FLOODrisk 2016 - 3rd European Conference on Flood Risk Management. DOI: 10.1051/ e3sconf/2016 0703019. Uncertainty and Expert Assessment for Supporting Evaluation of Levees Safety 1,a. 1,b. 2. 2. Michel Hathout , Marc Vuillet , Laurent Peyras , Claudio Carvajal and Youssef Diab. 1. 1.

(2) , 80 rue Rebeval, 75019 Paris, France Irstea, Research Unit RECOVER 3275 route Cézanne-CS 40064 F-13182 Aix-en-Provence Cedex 5, France. 2. Abstract. In France, levees remain most of the time badly maintained; these long linear structures show signs of weaknesses on numerous occasions. Only incomplete information is usually available. The general lack of data describing the behavior of the infrastructure during unwanted events led to estimate their safety mainly from expert judgment. Thus the ability of the expert to predict the level of functioning of an infrastructure for a type of hazard and its intensity is crucial. An error of judgment can have very serious consequences and the production of reliable information requires the ability of the expert to report accurately the uncertainties in its estimations, as well as associated confidence. In order to meet this need, our research within Incertu project (French Ministry of Ecology funding) aims to produce relevant scientific approaches and tools for the collection and processing reliable experts statements or combined with a confidence level in the context of uncertain information and input data.. 1 Introduction Levees of protection against flooding have a major role in the security of property and people. Unfortunately, disasters, those occur regularly in France and abroad, demonstrate whenever these structures are not infallible. The brutal characteristic of hazard aggravates the impact of flooding on people and other infrastructures. In France, the park of levees remains unclear. The structures are poorly documented and data acquisition across the linearity is complex and expensive, due to the age and heterogeneity of most levees. Currently, the evaluation of the performances of these structures is made with expertise, based on knowledges about their structural condition and on hazard that could be affected. In practice [1], the engineer / expert evaluates qualitatively the performance of structures, usually without an explicit formalization of his reasoning. Considering challenges associated with these structures, French regulations, via risk studies, now request the probabilistic risk assessment of failure of these structures for several scenarios of flooding. Therefore the engineer must make a subjective assessment of failure probability of the structure, in the absence of complete physical model and absence of enough statistical series to be usable in terms of reliability. However, the direct expression of the numeric data by the expert can sometimes lead to biased values [2, 3, 4]. These biases may be related to different factors. The judgments made by individuals based on heuristics [2], a b. that means informal reasoning, allowing the expert to quickly arrive at an evaluation or estimate, without the need of being really optimal. In a given situation, and to answer a specific question, the information by an expert can be a major source of information, but on the other hand it also leads to the necessity of developing approaches for the identification of bias, and if possible, to reduce them. Several approaches are described in literature concerning the familiarization with expert handling of probabilities in specific steps of elicitation or implementation of calibration procedures experts [5, 6]. Our scientific goal is to help identifying and reducing the biases that may affect the expert assessments of levees safety. We propose to answer to many scientific questions: - What are the biases that can occur during the expert evaluation of levees safety level? - How to apply a calibration approach as proposed by Cooke and Goossens [6] in the context of levees, and what would these contributions be? The studies of Vuillet [12] and Vuillet et al [1] propose some procedures to reduce bias that can occur when assessing the performance of levees. They are the starting point of our research in this project. Also, we present a brief summary of previous studies on the assessment of the safety level of operation of river levees and data imperfections typologies used to assess the safety of river levees. We also present a short summary of previous studies on the assessment of the safety level of operation of river levees and the typologies of. Corresponding author: michel.hathout@eivp-paris.fr Corresponding author: marc.vuillet@eivp-paris.fr. © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/)..

(3) E3S Web of Conferences 7, 03019 (2016) FLOODrisk 2016 - 3rd European Conference on Flood Risk Management. DOI: 10.1051/ e3sconf/2016 0703019. uncertainties in probabilistic format, it providing support to engineers for the diagnosis of structures for the main known failure scenarios. This model is intended to be used by a specialist engineer in rapid diagnostic status, by valuing all available data. The model provides support to the engineer for: identifying homogeneous portions of a linear of levees, to evaluate the performance of structures for different failure mechanisms and to specify the level of uncertainty of the produced results based on imperfect data available. Research process has three steps: (1) development of a functional model of the levees failure mechanisms, built from methods from Dependability and Qualitative Reasoning; (2) development of a support model deterministic decision including performance indicators for each failure mechanism levees, according to a method of construction of unique synthesis criteria; (3) development of a probabilistic performance assessment model including a method for taking into consideration the uncertainties of input information and results of the model in the context of subjective probabilities. In this approach, the engineer intervenes to assess the implementation of the functions assigned to the components levees on a semi quantitative performance scale from 0 to 10. He gives his most likely evaluation together with a range of uncertainty, as a modal value and interquartile interval [5%; 95%], that allow to construct probability density functions that can be successively spread through Monte Carlo simulation until a level of performance of the segments considered accompanied by margins of uncertainty. The approach also provides a path on processing through heuristics, mainly via a particular elicitation process and consulting a manual for handling subjective probabilities. In this state, these remain partial and still not possible to give some concrete elements to the biased nature of the valuations of an expert considering a particular area. Also, we present in the following sections the biases identified as likely to occur in the evaluation of the safety of levees and contribution of Cooke calibration model for quantified characterization of experts through a field of interest.. imperfections of data used to assess the safety of river levees. Then, we present a summary of experts' bias and approaches described in the literature proposing to reduce them. present the application of the calibration model of Cooke [5] in the field of levees by interviewing experts on the permeability of various commonly constituent materials for embankments.. 2 Methodologies for Assessment the Safety of Levees and Work of Vuillet et al Several approaches exist for assessing the safety of levees. Fauchard and Mériaux [16] present a synthesis of classical steps constituting levees diagnosis: preliminary studies including analysis of the levees history, topographical surveys, and detailed visual inspection. Geophysical and geotechnical knowledges are used to estimate the internal characteristics of the levees by studying variations of a physical field measured by longitudinal profiles. Numerical modeling is needed to calculate levees safety coefficients for certain failure mechanisms for profiles being the subject of substantial geotechnical knowledges. The qualitative assessment of levees by homogenous sections comes true by cutting linear units of incorporation and homogeneous loading. These diagnostic methodologies used primarily qualitative and do not take into account uncertainties remained of used data, nor aware of their impact on the determined safety level. The assessment data of the security level of levees are subjected to many imperfections: uncertainties of longitudinal and transverse representativeness of an incomplete survey in the absence of observation levees in flood, etc. These imperfections, combined with the variability of the phenomena of levees degradation, prevent the vast majority of cases the engineer / expert to commit a precise estimate of the actual behavior of the structure subjected to more or less strong hydraulic solicitation. The term of imperfection generally describes the concepts of quality and quantity of available data. Typically there are three main types of data imperfections: uncertainty, vagueness and incompleteness [17]. In our context of levees, the uncertainty can be a random or an epistemic one [4]. Random uncertainties are related to natural variability of material properties, epistemic ones are related to the dissemination of investigations along linear and representativeness of the available data [1]. Inaccuracies are related to the measures of physical phenomena, conditions and format for the visual inspection of the structures. The inaccuracies of investigative materials or assumptions are based on geotechnical testing [1]. Incompleteness instead, is related to the fact that levees are still generally poorly known structures. The lack of construction plan, the rarity of critical flood and lack of work monitoring, and the absence of the levee manager recognizing lead in general to deficit of information. Vuillet et al [1] propose a model for evaluating the performance of levees taking into consideration the. Figure 1. Uncertainty propagation criteria by Monte Carlo simulations for internal erosion performance indicator [1].. 2.

(4) E3S Web of Conferences 7, 03019 (2016) FLOODrisk 2016 - 3rd European Conference on Flood Risk Management. DOI: 10.1051/ e3sconf/2016 0703019.

(5) $

(6) # ,# "

(7) .

(8) -4 "# " "

(9) .

(10)

(11)

(12) "

(13) #

(14) 1

(15)

(16) " %

(17) #"

(18) " "

(19)

(20) "" #

(21) ,:- ! # 4 ". $

(22) $ "

(23) "

(24)

(25)

(26)

(27)

(28) ,- 1" " # 4

(29) *

(30) *

(31)

(32)

(33)

(34) "

(35) "

(36)

(37) 9 "

(38) . $ #

(39) 7 :"

(40)

(41) 4$ #

(42) "

(43) $<"

(44) ""

(45) . 3 Judgment Experts and Bias .

(46)

(47) .

(48)

(49)

(50)

(51)

(52)

(53)

(54) !

(55) "#

(56)

(57) $

(58) # #

(59)

(60) ". " "

(61) "

(62)

(63) %

(64)

(65)

(66) " %

(67)

(68)

(69) & !""

(70)

(71) '

(72)

(73) % ( &!

(74)

(75)

(76) %

(77)

(78) " "

(79) ) *.

(80) *

(81) #

(82)

(83) "

(84) " & +

(85)

(86)

(87) ". " % # $ #

(88) " % # ,

(89) -

(90) %

(91) "

(92)

(93)

(94) .

(95)

(96)

(97) $

(98)

(99)

(100) %

(101) " %

(102) #$

(103) %. 1$ # #

(104) " "

(105)

(106)

(107) " "

(108)

(109) $ .

(110) " #

(111)

(112) &$ # &

(113) !$ # # "

(114)

(115)

(116)

(117)

(118)

(119)

(120) & &

(121) =4

(122) $

(123)

(124) "

(125) "

(126) " ,

(127) $

(128) # " -

(129)

(130) " >. !

(131)

(132)

(133) / &

(134)

(135)

(136)

(137) #

(138) %

(139) "

(140)

(141) *

(142) & (

(143) ". " /

(144) %

(145)

(146) *

(147) 0 + / % #"% " ,

(148) *

(149) * -

(150)

(151) 1

(152)

(153)

(154) "

(155)

(156) ## "

(157) " 2

(158) "

(159)

(160) "

(161) 3 + " "

(162)

(163) "

(164) %

(165)

(166) #/" $" 4

(167)

(168)

(169) / #

(170)

(171) $ ,

(172) -. 1

(173) #

(174)

(175)

(176)

(177) "

(178)

(179) " #

(180)

(181) 7 1

(182) " "

(183)

(184)

(185) "

(186) .

(187) "

(188)

(189) ## =4 " "

(190) "

(191) %

(192)

(193) " . " "

(194) $ # #

(195)

(196)

(197)

(198) 1

(199) $"

(200) " #

(201) ) "

(202) ;

(203) $

(204) $ " .

(205) $ #

(206) =4

(207) $"

(208) $

(209)

(210) ) "

(211) "

(212) $ # $ #

(213) . 5 % %

(214) "

(215)

(216) #

(217)

(218)

(219)

(220) ## "/ 6+

(221)

(222) " " # %

(223)

(224) 1 " # "

(225) #

(226) ""

(227) "

(228)

(229) " "

(230)

(231) % 7 1 "

(232) "

(233)

(234)

(235) %

(236)

(237) )

(238)

(239) "

(240) #

(241) $ 2

(242) 8. 1"

(243) #

(244) "" $

(245)

(246) " "

(247)

(248) $

(249)

(250)

(251) # " /

(252) / =4 "

(253)

(254)

(255) " $

(256) " $ "

(257) ""

(258)

(259)

(260) %

(261) /

(262) /

(263)

(264)

(265) "

(266) " . 1 #/" $ $

(267)

(268)

(269) " "

(270) "

(271)

(272) "

(273)

(274) $ " &9

(275)

(276)

(277) . . "

(278)

(279)

(280) #

(281) $ * "%#

(282) "

(283) " & : 1 # &5

(284)

(285) #

(286)

(287)

(288)

(289)

(290) # $

(291)

(292) " "

(293) "

(294) & : 1 ;

(295) . $ %

(296) # " # + "

(297) " %

(298)

(299) # %

(300)

(301) " 2

(302)

(303) 7. 4 Debiasing Approaches ?$

(304) " $ # $

(305)

(306) "

(307)

(308) "

(309)

(310)

(311) # 4

(312)

(313) $

(314) #

(315) #

(316) @

(317) ". " "

(318)

(319) " : )

(320)

(321) " @ 9 $

(322)

(323)

(324) $

(325)

(326) "

(327)

(328)

(329) A

(330)

(331) "

(332) #

(333)

(334) $

(335)

(336) "

(337) "

(338)

(339) " "

(340)

(341) #/" $

(342) ## . . 1

(343) "

(344)

(345) # ( ,- ;

(346) . $ # ,#

(347) -4

(348) "

(349)

(350) "

(351)

(352) %

(353) "

(354) #

(355)

(356) "

(357) ,7-. B

(358) "

(359) #

(360) .

(361) $ $

(362)

(363) . 3.

(364) E3S Web of Conferences 7, 03019 (2016) FLOODrisk 2016 - 3rd European Conference on Flood Risk Management. ""

(365)

(366) ; < *

(367)

(368) #

(369) "

(370) ## $"

(371) " #

(372) $"

(373)

(374)

(375) # $"

(376)

(377) #/"

(378) ## $"

(379)

(380)

(381) "

(382) ## . " #

(383)

(384) C

(385) C.

(386)

(387)

(388) $ #

(389) %

(390) ," #

(391) $ # -1" #

(392)

(393) "

(394)

(395) "

(396) " #

(397)

(398)

(399) " # %

(400) $ , 2

(401) # $-

(402)

(403) $ "

(404) " # $ " # %

(405) $ "

(406)

(407) " #

(408) 81 $

(409)

(410) " #

(411) 4 " "

(412)

(413) ;

(414)

(415) %

(416) &;

(417)

(418) %& '

(419)

(420) <?

(421) &?&

(422) & &

(423) D

(424)

(425)

(426) " "

(427)

(428) ;

(429)

(430) % "

(431) ##

(432) "

(433)

(434)

(435) @ + ". 2 "

(436) . $#

(437) ". 5 Application of the Classical Model of Cooke on the Levees + "

(438) " " " #

(439)

(440)

(441) ;

(442)

(443) %

(444)

(445)

(446) " #

(447) $ #

(448) " "

(449) "

(450) #

(451) "

(452) " $

(453) $ #

(454) ' " #

(455) "" ) C

(456)

(457) %

(458) #

(459) $ #

(460) ,

(461)

(462)

(463) $

(464)

(465)

(466) " " $ " - 7 ?

(467) #% $ "

(468)

(469)

(470)

(471)

(472)

(473) "

(474) " $ 1

(475) %# $

(476) "

(477)

(478)

(479) ,6<0 2 - ,6<@ 2 -

(480) 8 1 "

(481) $ 4 - <

(482)

(483) " " 7-<

(484) $ " :-< /

(485) " $ -</

(486) 8-</

(487)

(488) " " @-< /

(489) $ " (-<1

(490) $1

(491) #/" $. " $

(492) " #

(493) $ #

(494) " "

(495) # "

(496) "

(497) #

(498) ,6<7 2 - ,!

(499)

(500)

(501) - ,D # -. ,7-4 . 1" "

(502)

(503) ;

(504)

(505) %8

(506) " "

(507) #

(508)

(509)

(510) $

(511)

(512) $ 1 ""

(513) #

(514) #

(515) " #

(516) "

(517) * *

(518)

(519) ## "

(520) 1 " #

(521) #

(522) $

(523)

(524) "

(525) " # $ " # %

(526) $ 1

(527)

(528) # $

(529)

(530) ," #

(531) "

(532) -

(533)

(534) %

(535) " 1" #

(536) "

(537) "

(538)

(539) "

(540)

(541) 4 "

(542)

(543)

(544) " #

(545) "

(546)

(547)

(548)

(549) 1 " #

(550) "

(551)

(552) *

(553) .

(554)

(555) $# + #

(556) " #

(557) "

(558) ## $"

(559) 6 " #

(560) $"

(561)

(562) #/" $

(563) ## $"

(564) , # - 1 "

(565)

(566) $ " #

(567) "

(568) %

(569) $

(570) $ #

(571) " #

(572)

(573) . $ " # 1

(574) "

(575)

(576) *

(577)

(578)

(579) $# .

(580)

(581) #

(582) # "

(583)

(584) ## "

(585) + #

(586)

(587) " # #/" $

(588) ## $"

(589) , $ # -

(590) "

(591) ## $"

(592)

(593) "

(594) ## "

(595) 1 "

(596)

(597) $

(598) "

(599)

(600) " #

(601)

(602) . $

(603) " " #

(604)

(605) #

(606) 1 ""

(607)

(608)

(609) "

(610)

(611) #

(612)

(613)

(614)

(615) "

(616)

(617) &" " "&4 &9 .

(618)

(619) & #

(620) $"

(621) " &" "&

(622)

(623)

(624) . Ce. 1 & 2 2 n I e ci , pt

(625)

(626) . ,-. 1 z ¦ I e pt , s i

(627) n 1. ,7-. Ke. DOI: 10.1051/ e3sconf/2016 0703019. ?"

(628) # "

(629) " ,2 -. 1

(630) $

(631)

(632) < *

(633) $ "

(634) " # 6E 36E $ "

(635)

(636) %86E ,:- #

(637) #/" $

(638) ## #

(639)

(640) " $

(641) " #

(642) $ #4 "

(643) "

(644) #

(645) +

(646) "

(647) %

(648) $

(649) " #

(650) $ # ,9-

(651) "

(652)

(653) " #

(654) $ # " "

(655)

(656) "

(657) " #

(658) $ # #

(659)

(660) : . 4 $"

(661)

(662) #/" $

(663) ## . $"

(664)

(665)

(666) "

(667) ## $"

(668)

(669) " #

(670) e # #

(671) <* "

(672) n #

(673) " #

(674) $ # #

(675) "

(676)

(677) $ # . 4. .

(678) E3S Web of Conferences 7, 03019 (2016) FLOODrisk 2016 - 3rd European Conference on Flood Risk Management. DOI: 10.1051/ e3sconf/2016 0703019. 1

(679) " #

(680)

(681) 7:G8# "

(682)

(683) "

(684)

(685) " #

(686) @G(

(687) # $ ,7- # " #

(688) "

(689) , $ ,- ,:- ,- ,8- 1 ,@- ,(- $

(690) " #

(691) "

(692) , $ +#

(693) " $

(694) " #

(695) $ #

(696)

(697)

(698) " $ " # . 1 #/" $

(699) ## #

(700)

(701) " $

(702) " #

(703) $ #4 "

(704) "

(705) #

(706) <" . . 1

(707)

(708) "" )

(709)

(710)

(711) "

(712) "

(713) " #

(714)

(715) #/" $

(716) ## #

(717) " # !

(718) "" " #

(719) "

(720)

(721) "

(722) " " #

(723) $"

(724)

(725) "

(726)

(727)

(728) "" #

(729) $ #""

(730) "

(731)

(732)

(733) $"

(734)

(735)

(736) "

(737) ## $

(738) " #

(739) $"

(740) % #

(741)

(742) "

(743) ## $"

(744)

(745) " # <* " 6E. 6E. 86E. 36E. 66E #

(746) " #

(747) $ #

(748)

(749)

(750) "<" # . $

(751) " #

(752) $ #

(753) " " 6E.36E 06E

(754) " 76E

(755) $ * # $ 6. 6E 36E. 66E ! "

(756) ""

(757) "

(758)

(759) " #

(760) "

(761)

(762) # ""

(763) " $.

(764)

(765)

(766) #/" $

(767) ## $"

(768) $

(769)

(770) "

(771) ## $"

(772)

(773) "

(774)

(775) ͛ #

(776) "

(777)

(778) $ # $

(779)

(780) "" ! $

(781) # #

(782)

(783) 1# ,-

(784) " #

(785) . 1 " #

(786)

(787)

(788) $ $ "

(789) 2

(790) $

(791) #

(792) H $

(793)

(794) " #

(795) $ #, - !

(796) 2 $ "

(797) " "

(798)

(799) + .

(800)

(801)

(802) $ $ "

(803)

(804)

(805) " #

(806) *

(807)

(808)

(809)

(810) ## "

(811)

(812)

(813) """ A

(814) "

(815)

(816)

(817) %

(818) #

(819)

(820)

(821)

(822)

(823) "

(824)

(825) ## "

(826)

(827) C #

(828) #"

(829) # $

(830) """

(831) .

(832)

(833) $ # "

(834) $ #, 6<0-

(835)

(836)

(837) $ " " # " # " " $

(838) " , " "

(839) $ # -1

(840)

(841)

(842) "

(843) "

(844) " #

(845) $ #

(846) $ #

(847)

(848)

(849)

(850) "

(851) " # #

(852)

(853)

(854) " #

(855) $

(856)

(857) $ 1 $

(858) #" $ " " *

(859) " . #

(860) %

(861)

(862) $

(863) # 2 "

(864) " $ #

(865)

(866) "

(867) . " #

(868) " #

(869) $ #

(870)

(871) $

(872)

(873) "

(874) ## $"

(875)

(876) # " # 1 $ "

(877)

(878)

(879) " #

(880)

(881)

(882)

(883)

(884) # "

(885)

(886) #

(887) C

(888) " #

(889) $ # " "

(890) $.

(891)

(892) #

(893) #

(894) "

(895) " #

(896) $ # . . . ͛F . F͛ ͛. = . 0.42. 11.9. 5.04. 0.26. 6 Continuity and Application. = 7. 0.42. 12.7. 5.36. 0.28. = :. 0.42. 11.2. 4.75. 0.25. = . 0.42. 8.5. 3.59. 0.19. = 8. 0.06. 4.4. 0.25. 0.013. = @. 0.004. 6.5. 0.03. 0.001. = (. 0.004. 16.9. 0.07. 0.004. 1 " $

(897) +" "

(898) /" 76:<76@, # == "

(899) 9 %5".

(900) 1

(901) 76:-

(902) " =+AI+ J $ I =

(903) " 5

(904) D

(905) $ ,5A5-

(906) ;

(907) I + K5 5"<9

(908) " <+5+59 !

(909) "

(910) -. " 766%

(911) $

(912) L

(913) #

(914)

(915)

(916) "

(917) $

(918) ## .

(919)

(920) $ 1

(921) "

(922) . #

(923) " ) $

(924)

(925)

(926) " "

(927) + ! I

(928) $" !5<+591

(929)

(930)

(931)

(932) "