* The figure indicates that if the average of b s /a s values is less than 0.5, the particles are an oblate ellipsoid. Figure 9 shows the variation of N V with time in a system in which DNϭ100 particles are nucleated successively per unit time and grow at the same rate as in Fig. 8. The rate of increase in the particle number was determined to be J V ϭ 102.8 by least square method, close to the increasing rate of actual particle numbers. These figures show that the procedures described above can correctly estimate the particle density in the case of both constant and varying particle numbers. Figure 10 shows the variation of N V with time calculated with the same nucleation and the growth rate as in Figs. 8 and 9, incorporating particle impingement. The N V first increases with time, but turns to decrease at tϳ8. If nucleation stops at tϭ8, the amount of decrease becomes greater. Figure 11 shows the variation with time of the number of clusters of impinging particles, where n is the number of sections contained in one cluster. Solid triangles are the sum of the clusters of various sizes to be observed on the test plane. Thus, the influence of particle impingement on the estimation of particle density is very large. The simulat- ed temporal variation of N V is similar to that of the number of ferrite allotriomorphs in Fe-C(-X) alloys reported previously, 15,16) which indicates that impingement of ferrite particles already occurred within austenite grain boundaries, although the fraction transformed was small at the time of measurement.
Simulation Incorporating Particle ImpingementThe omission of undetectably small particles can have a large influence on the estimation of the total number of particles since the weight factor a i is large for small i ( Table 1). The amount of error may depend on the growth kinetics, that is, if the initial growth rate of particles is large the proportion of small particles is small.
Determination of the Number of Non-sphericalParticles The N V of Nϭ500 oblate particles of qϭ0.4 nucleated under the site saturation mode was calculated and results are shown in Fig. 12a. Solid symbols were calculated from Eq. (13). They indicate that N V is correctly evaluated by measuring oblate particles along the major axis of elliptical sections if the shape factor (ϭ0.832, Fig. 3) is included. If these particles are approximated as spheres, N V is underestimated as shown by open symbols. The amount of error is consistent with the omission of the shape factor.On the other hand, prolate particles are classified by measuring the length along the minor axes of elliptical sections.Simulation results for prolate particles of qϭ3 are shown in Fig. 12b. The N V is again correctly evaluated by incorporating the shape factor (ϭ1.81) as shown by solid symbols. A slightly smaller N V value at the earliest time (tϭ0) is probably because D max (ϭ20) was too large compared to the particle size at that time (2rϭ1). If these particles were treated as spheres and the particle size was measured along the major ...