The accuracy of core samples as measures of lean content in cartons of beef: A simulation study

Cartons of boneless beef with dimensions of 500 mm × 300 mm × 200 mm were simulated with 48,000 elements of 5 mm × 5 mm × 25 mm arranged in 8 layers each 25 mm thick. Each element was randomly allocated to be either lean or fat in such a way that parameters in the randomisation process could be varied so that the lean percentage was varied. In addition, sizes of lean and fat pieces were varied by specifying the frequency with which adjacent elements within a layer were the same, both along rows in the 500 mm dimension, and across rows. The composition of material from 12 core samples was then compared with total carton composition. The primary measure of coring accuracy used was the standard deviation of differences between total lean% and core lean% when runs of 30 cartons were simulated. These were less than 1 when samples of lean and fat within a layer were the size of single elements and the total lean% was 80 or more, but increased to more than 2 when samples sizes were larger or total lean% was lower. The residual standard deviations from regressions of total lean% on core lean% were seldom less than 2. Using this approach, quantitative estimates showed that coring accuracy improved with increasing lean% levels, with decreasing size of lean and fat pieces, and with increasing core size. However, the study considered only one step in the sampling process, with overall accuracy also being dependent on the extent to which the cartons sampled are representative of the whole load, and the accuracy with which fat and lean content of the core sample is measured. Overall the study suggested that the accuracy of estimation of carton chemical lean content from standard coring procedures is not particularly high. © 2007 Elsevier B.V. All rights reserved.