Calculations involving a single random variable (SRV)

Similar documents
Prandl established a universal velocity profile for flow parallel to the bed given by

Effects of Soil Spatial Variability on Bearing Capacity of Shallow Foundations

UNCERTAINTY FOCUSED STRENGTH ANALYSIS MODEL

An Investigation into Estimating Type B Degrees of Freedom

Formal Methods for Deriving Element Equations

DISPLACEMENT ANALYSIS OF SUBMARINE SLOPES USING ENHANCED NEWMARK METHOD

Lecture Notes: Finite Element Analysis, J.E. Akin, Rice University

08.06 Shooting Method for Ordinary Differential Equations

Workshop on Understanding and Evaluating Radioanalytical Measurement Uncertainty November 2007

BEHAVIOUR OF LATERALLY LOADED RIGID PILES IN COHESIVE SOILS BASED ON KINEMATIC APPROACH

Simplified Identification Scheme for Structures on a Flexible Base

Sources of Non Stationarity in the Semivariogram

Technical Note. ODiSI-B Sensor Strain Gage Factor Uncertainty

Applying Laminar and Turbulent Flow and measuring Velocity Profile Using MATLAB

Integration of Basic Functions. Session 7 : 9/23 1

Chapter 1: Differential Form of Basic Equations

Kragujevac J. Sci. 34 (2012) UDC 532.5: :537.63

Curves - Foundation of Free-form Surfaces

Section 7.4: Integration of Rational Functions by Partial Fractions

Study on the impulsive pressure of tank oscillating by force towards multiple degrees of freedom

Chapter 2 Introduction to the Stiffness (Displacement) Method. The Stiffness (Displacement) Method

EDEXCEL NATIONAL CERTIFICATE/DIPLOMA. PRINCIPLES AND APPLICATIONS of FLUID MECHANICS UNIT 13 NQF LEVEL 3 OUTCOME 3 - HYDRODYNAMICS

i=1 y i 1fd i = dg= P N i=1 1fd i = dg.

Steady State and Transient Thermal Analysis of Switched Reluctance Machine

Uncertainty Analysis of the Thunder Scientific Model 1200 Two-Pressure Humidity Generator

Approximate Solution for the System of Non-linear Volterra Integral Equations of the Second Kind by using Block-by-block Method

Material Transport with Air Jet

Second-Order Wave Equation

Principles of Minimum Cost Refining for Optimum Linerboard Strength

Efficiency Increase and Input Power Decrease of Converted Prototype Pump Performance

1 Differential Equations for Solid Mechanics

3 2D Elastostatic Problems in Cartesian Coordinates

5. The Bernoulli Equation

EOQ Problem Well-Posedness: an Alternative Approach Establishing Sufficient Conditions

An Introduction to Geostatistics

Appendix A: The Fully Developed Velocity Profile for Turbulent Duct Flows

ON THE SHAPES OF BILATERAL GAMMA DENSITIES

Setting The K Value And Polarization Mode Of The Delta Undulator

A Single Species in One Spatial Dimension

DESIGN OF STRAP (CANTILEVER) FOOTINGS Design Steps and Equations

Thermal balance of a wall with PCM-enhanced thermal insulation

Limit-Equilibrium Stability Analysis of Spiling Soil Reinforcement in Tunneling

Reflections on a mismatched transmission line Reflections.doc (4/1/00) Introduction The transmission line equations are given by

Lateral Load Capacity of Piles

Determining of temperature field in a L-shaped domain

Incompressible Viscoelastic Flow of a Generalised Oldroyed-B Fluid through Porous Medium between Two Infinite Parallel Plates in a Rotating System

A New Approach to Direct Sequential Simulation that Accounts for the Proportional Effect: Direct Lognormal Simulation

Design and Data Acquisition for Thermal Conductivity Matric Suction Sensors

Cosmic Microwave Background Radiation. Carl W. Akerlof April 7, 2013

METHODOLOGY FOR EXPERIMENTALLY DETERMINING THE CHARACTERISTICS OF MEDIUM VOLTAGE ZINC OXIDE VARISTORS

Restricted Three-Body Problem in Different Coordinate Systems

MEASUREMENT OF TURBULENCE STATISTICS USING HOT WIRE ANEMOMETRY

A Regulator for Continuous Sedimentation in Ideal Clarifier-Thickener Units

Chapter 3 MATHEMATICAL MODELING OF DYNAMIC SYSTEMS

RELIABILITY-BASED DESIGN INCORPORATING MODEL UNCERTAINTIES. K. K. Phoon Department of Civil Engineering, National University of Singapore, Singapore

Inertial Instability of Arbitrarily Meandering Currents Governed by the Eccentrically Cyclogeostrophic Equation

Propagation of measurement uncertainty in spatial characterisation of recreational fishing catch rates using logistic transform indicator kriging

Chapter (11) Pile Foundations

Evaluation of the Fiberglass-Reinforced Plastics Interfacial Behavior by using Ultrasonic Wave Propagation Method

Chapter 6 Momentum Transfer in an External Laminar Boundary Layer

Uncertainty Evaluation of Toluene Determination in Room Air by Thermal Desorption Gas Chromatography

Partial Differential Equations with Applications

Copyright Canadian Institute of Steel Construction

Stability of structures FE-based stability analysis

ECON3120/4120 Mathematics 2, spring 2009

OPTIMUM EXPRESSION FOR COMPUTATION OF THE GRAVITY FIELD OF A POLYHEDRAL BODY WITH LINEARLY INCREASING DENSITY 1

Nonlinear parametric optimization using cylindrical algebraic decomposition

The Linear Quadratic Regulator

Generalized Jinc functions and their application to focusing and diffraction of circular apertures

Momentum Equation. Necessary because body is not made up of a fixed assembly of particles Its volume is the same however Imaginary

Transient Approach to Radiative Heat Transfer Free Convection Flow with Ramped Wall Temperature

The Dual of the Maximum Likelihood Method

Math 116 First Midterm October 14, 2009

FOUNTAIN codes [3], [4] provide an efficient solution

FEA Solution Procedure

Bayes and Naïve Bayes Classifiers CS434

4 Exact laminar boundary layer solutions

FLUID FLOW FOR CHEMICAL ENGINEERING

All India Mock GATE Test Series Test series 4 Civil Engineering


SOIL NON-LINEAR BEHAVIOR AND HYSTERETIC DAMPING IN THE SPRING-DASHPOT ANALOG

Geometric Image Manipulation. Lecture #4 Wednesday, January 24, 2018

Frequency Estimation, Multiple Stationary Nonsinusoidal Resonances With Trend 1

Chapter 2 Difficulties associated with corners

THE EFFECTS OF RADIATION ON UNSTEADY MHD CONVECTIVE HEAT TRANSFER PAST A SEMI-INFINITE VERTICAL POROUS MOVING SURFACE WITH VARIABLE SUCTION

2E1252 Control Theory and Practice

Localization in Undrained Deformation 1

BLOOM S TAXONOMY. Following Bloom s Taxonomy to Assess Students

called the potential flow, and function φ is called the velocity potential.

Two identical, flat, square plates are immersed in the flow with velocity U. Compare the drag forces experienced by the SHADED areas.

Modeling Effort on Chamber Clearing for IFE Liquid Chambers at UCLA

DILUTE GAS-LIQUID FLOWS WITH LIQUID FILMS ON WALLS

Microscopic Properties of Gases

ρ u = u. (1) w z will become certain time, and at a certain point in space, the value of

PhysicsAndMathsTutor.com

Consistent Numerical Model for Wind Buffeting Analysis of Long-Span Bridges

A fundamental inverse problem in geosciences

EXCITATION RATE COEFFICIENTS OF MOLYBDENUM ATOM AND IONS IN ASTROPHYSICAL PLASMA AS A FUNCTION OF ELECTRON TEMPERATURE

Department of Industrial Engineering Statistical Quality Control presented by Dr. Eng. Abed Schokry

Transcription:

Calclations involving a single random variable (SRV) Example of Bearing Capacity q φ = 0 µ σ c c = 100kN/m = 50kN/m ndrained shear strength parameters What is the relationship between the Factor of Safety (FS) of a conventional bearing capacity calclation (based on the mean strength) and the probability of failre (p f )?

First perform a deterministic calclation Conventional bearing capacity calclations typically involve high factors of safety of at least 3. The bearing capacity of an ndrained clay is given by the Prandtl eqation: q = ( +π ) c = 5.14c where c is a design mean vale of the ndrained shear strength. If c = 100kN/m, and FS = 3 this implies an allowable bearing pressre of: q all 5.14 100 171kN/m = = 3

Now perform a probabilistic calclation If additional data comes in to indicate that the same ndrained clay has a mean shear strength of µ = 100kN/m and a standard deviation of σ = 50kN/m c c and is lognormally distribted, what is the probability of bearing failre? In other words, what is the probability of the actal bearing capacity being less than the factored deterministic vale Pq [ < 171]? q = 5.14 c, hence if c is a random variable we can write: E[ q ] = 5.14E[ c ] ths µ =5.14 µ =514 q c and Var[ q] = 5.14 Var[ c] ths q =5.14 c =57 σ σ (Note that since q c, V = V q c = 1 ) Failre occrs if q < q all The probability of this happening can be written as Pq [ < 171]

First find the properties of the nderlying normal distribtion of ln q { } 1 σ ln q = ln 1+ V ln 1 0.47 q = + = 1 1 µ ln q = ln µ ln ln(514) (0.47) 6.13 q σ q = = ln171 6.13 Pq [ < 171] =Φ 0.47 =Φ.10 ( ) = 1 Φ.10 ( ) = 1 0.98 = 0.018 (1.8%)

Example of Slope Stability φ = 0 µ, σ γ c sat c Dimensionless strength parameter C = c γ H sat What is the relationship between the Factor of Safety (FS) of a ope (based on the mean strength) and the probability of failre (p f ) of an ndrained clay ope for different vales of the mean and standard deviation of strength?

First perform a deterministic calclation Using limit eqilibrim or charts, find the FS corresponding to a niform ope of strength µ c for a homogeneos ope FS α C Linear relationship between C and FS for a cohesive ope with a ope angle of β =6.57 and a depth ratio of D = o

Now perform a probabilistic calclation 1) Noting that FS C or FS = KC, compte K, hence µ = Kµ and V = V FS C FS C ) If C is lognormal, so is FS, so compte the nderlying normal properties σ and µ 3) Find the probability P[ FS < 1 ] from standard tables. ln1 µ ln FS µ ln FS If FS is lognormal, P[ FS < 1] =Φ =Φ σ σ ln FS ln FS ln FS 1 µ FS If FS is normal, P[ FS < 1] =Φ σ FS ln FS

Single Random Variable (SRV) approach Defining FS 5.88µ C Probability of Failre p f or p f = p( FS < 1) = p( C < 0.17) High! V C σ C = µ C The mean strength may be too optiminstic. Probability of failre p vs. Factor of Safety FS (based on the mean) f

The median helps to interpret lognormal behavior p f MedianC < 0.17 MedianC > 0.17 p vs. V for different vales of Median f C C

Mean strength factoring All crves assme FS=1.47 based on C des = 0.5 linear scaling C = µ f µ des C 1 C C sd scaling = µ f σ des C C

All the examples considered so far were for a single random variable (sally the ndrained strength) How can we accont for more than one random variable? (e.g. the cohesion and friction angle in a drained analysis) A well-established approximate method for estimating the inflence of several random inpt variables on a fnction is the First Order Second Moment (FOSM) method. The method is called First Order becase it incldes only first order terms in a Taylor expansion. The method is called Second Moment becase it enables estimates to be made of the variance (second moment) of fnction nder consideration.

Earth Pressre Example sing FOSM with 4 Random Variables 0.30 Cohesionless soil H = 5.03 4.57 c = 0, φ Active Earth Pressre Coefficient K a φ = tan (45 ) γ c γ bf P a Units in kn and m 0.46 1.37 1.83 tanδ 3.66

0.30 4.57 H = 5.03 Cohesionless soil c = 0, φ Active Earth Pressre Coefficient K a φ = tan (45 ) 1 P = γ H K = γ K Rankine's Theory: a bf a 1.65 bf a W = W + W γ c γ bf P a 1.37 1.83 0.46 tanδ 3.66 Factor of Safety against iding: c bf 1 = 0.46(3.66) γ + (4.57)(0.30 + 0.46) γ + 1.83(4.57) γ = 3.4γ + 8.36γ c c c bf bf FS = Units in kn and m ( c + bf ) 3.4γ 8.36γ tanδ 1.65γ bf K a

Inpt data assming all variables are random FS is a fnction of 4 random variables: FS = f ( γ c, γbf, Ka,tan δ) Property γ c (kn/m 3 ) µ 3.58 σ 0.31 γ bf (kn/m 3 ) 18.87 1.10 K a 0.333 0.033 tan δ 0.5 0.05

Now se the FOSM Method to estimate the statistics of FS ( c + bf ) 3.4 E[ γ ] 8.36 E[ γ ] E[ tan δ] E[ FS] = µ 1.65 E[ γ ] E[ K ] bf a FS FS FS FS FS Var[ FS] Var[ γ c] + Var[ γbf] + Var[ Ka] + Var[ tan δ] γ c γ bf Ka (tanδ) = σ FS The mean vale of can be estimated in the FOSM Method by sbstitting all the FS mean vales of the inpt variables into the governing fnction, ths: ( + ) 3.4(3.58) 8.36(18.87) 0.5 Hence, µ FS = 1.50 1.65(18.87)0.333 Since we have a fnctional relationship between FS and the inpt variables, the variance of FS can be estimated in this case by evalating the derivatives analytically at the mean.

Factor of Safety against iding: FS = ( c + bf ) 3.4γ 8.36γ tanδ 1.65γ bf K a FS γ 0.7 tanδ = = γ K c bf a 0.0 FS γ 0.7γ tanδ = = 0.03 c bf γbf Ka FS K (0.66γ + 0.7 γc ) tanδ = = 4.50 bf a γ bfka FS (0.66γ 0.7 ) bf + γ c = = 3.00 (tanδ) γ K bf a All derivatives evalated at the mean

Hence: Var[ FS ] 0.0 0.31 + 0.03 1.10 + 4.50 0.033 + 3 0.05 = 0.046 σ FS = 0.046=0.1 µ FS = 1.50 σ FS Smmary: VFS = = 0.14 σ FS = 0.1 µ FS What is the probability of iding failre? P[ FS < ] 1??

In order to compte probabilities we need to assme a sitable PDF Let s assme that FS is lognormally distribted First find the mean and standard deviation of the nderlying normal distribtion σ ln FS { } { V } = ln 1+ = ln 1+ 0.14 = 0.14 FS 1 1 µ ln FS = ln µ ln ln(1.5) (0.14) 0.40 FS σ FS = = [ 1] P FS ln1 0.40 < =Φ 0.14 =Φ(.86) = 1 Φ(.86) = 1 0.999781 = 0.000 or 0.0%

If we instead assme that FS is normally distribted [ 1] P FS 1 1.5 < =Φ 0.1 =Φ(.38) = 1 Φ(.38) = 1 0.991343 = 0.0087 or 0.9 %

Nmerical Differentiation In this case we were able to differentiate analytically, bt in some problems we do not have a fnction to work with (e.g. ope stability). In these case we can se nmerical differentiation Reworking of the earth pressre problem sing nmerical differentiation. FS = f ( γ, γ, K,tan δ) c bf a FS FS FS FS Var[ FS] Var[ γ c] + Var[ γbf] + Var[ Ka] + Var[ tan δ] γ c γ bf Ka (tanδ) We can evalate derivatives sing a central difference formla in which we sample the fnction at one standard deviation below and one standard deviation above the mean. e.g. FS K f( µ γ, µ γ, µ K + σ K, µ tanδ) f( µ γ, µ γ, µ K σ K, µ tanδ) fk = σ σ c bf a a c bf a a a a K K Note all variables are fixed at their mean except for the derivative variable a a

After sqaring the derivative, the standard deviations cancel and we are left with: fγ bf Ka tanδ fγ f f c Var[ FS] + + + Compte the vales sing a tablar approach Property γ c (kn/m 3 ) µ 3.58 σ 0.31 µ± σ 3.89 3.7 f 1.50 1.49 f 0.01 γ bf (kn/m 3 ) 18.87 1.10 19.97 17.77 1.47 1.53-0.06 K a 0.333 0.033 0.367 0.300 1.36 1.66-0.30 tan δ 0.5 0.05 0.55 0.45 1.65 1.35 0.30

Now sbstitte these vales into the variance eqations 0.01 0.06 0.3 0.3 Var[ FS] + + + = 0.046 Hence, σ FS = 0.1 which is the same vale (to decimal places) as the one we obtained analytically. The fact that the analytical and nmerical rets are very similar in this case indicates that there is almost no nonlinearity in this problem.

The Reliability Index The Reliability Index is a measre of the margin of safety in standard deviation nits. For example, if dealing with a Factor of Safety, the reliability Index is given by: β = µ FS σ 1 FS In a normal variate, the reliability index (β) is niqely related to the probability of failre (p f ) throgh the expression: p f = 1 Φ( β )

Consider a normal distribtion of the Factor of Safety (FS) f FS µ σ FS FS = 1.5 = 0.1 p f is this area FS β is this distance the standard deviation (1.5 1) β = =.38 p f = 1 Φ(.38) = 1 0.991343 = 0.0087 0.1

0.5 Probability of Failre: p f Reliability Index: β Probability of Failre vs. Reliability Index for a Normal Distribtion