−1.2 −1 −0.8 −0.6 −0.4 −0.2 −14 −12 −10 −8 −6 −4 −2 1 6 0 2 6 0 3 6 0 5 4 0:19 10 2:8 0:34 =1 2 3 5 10 Model 1 Model 2 n=15 15 n=25 25 n=30 30 log( = Æ) log( log(1 W( ))) Figure 2: Comparison of the weakest link distribution predicted by the 2D cluster growth model with the empirical weak link distributions ^W(x) obtained from Monte-Carlo simulations. Model 1 assumes =p4 and = 0:5 for all . In Model 2, we adjust the parameters and so as to get the best t with the simulated data. These values of and are listed in the bottom right corner. Weakest link distributions corresponding to = 0:5 are not shown because both Model 1 and Model 2 predict distributions that are out of the range of this plot. Also omitted is the Model 1 line for = 1 which also lies outside the range of this plot. where Kr is the stress concentration on the Nr most severely overloaded neighbors of a tight r-cluster and is given by Eqn. 2. We introduce two parameters, and , to account for the actual numberof bersathighriskoffailure.Let Nr = r (5) be the number of severely overloaded neighbors around anr-cluster, where >0and0 < 1=2. We ndthisstructurefor Nr to be essential in order that Wn( ) in Eqn. 4 agree with the form of the simulated ^W( ) distribution especially for small . Taking =p4 3:55 and =1=2, impliescountingallthe bersontheclusterperipherytobeatriskoffailure.Model1linesin Figure2correspondtothiscaseanddonot tthesimulated^W( ) very well. However model 2 lines in which we vary and as functions of t the simulated ^W( ) much better. A closed form approximation for W( ) is derived in [5]. 2. Large Variability in Fiber Strength (small ) When issmallcorrespondingtolargevariabilityin berstrength,thecluster-drivenbreakdown mechanism is dominated by a dispersed, strength-driven breakdown mechanism of the Æ-bundle. This is clearly seen in the failure snapshots of a Æ-bundle for = 1, as shown in Figure 3. Inthecaseofdispersed berfailureina Æ-bundle,thedetailsofthe berload-sharingmodelmay not be important provided that the model conserves load. Thus we consider behavior under the equal load-sharing rule or ELS. ELS assumes that the stress concentration factor for each intact ber in an n- ber Æ-bundle with j broken bersis n;j =n=(n j). Applying a result due to Smith[3]whichsharpensoneduetoDaniels[4]toanELSbundleofWeibull berswe ndthat the bundle strength distribution Gn( ) converges as n!1, to the normal form (( n)=s n)
RkJQdWJsaXNoZXIy MjM0NDE=