Modelling the size and shape of hoki for improved process control

Application and testing of a simple method for predicting chilling times for hoki (Macruronus novaezelandiae)

“…The cross-section of the fish is approximately elliptical, but the body is elongated with a long, thin tail. Bickers, Davies, and MacDonald (2000) measured and correlated dimensional data for spawning and non-spawning hoki. Their data indicate that hoki length is approximately 14 times the body thickness and 7.5 times the body depth.…”

“…To apply the chilling time prediction method to these trials, thickness and depth dimensions were estimated from fish length using the following correlations derived by Bickers et al (2000) for non-spawning hoki:…”

“…The dimensional data reported by Bickers et al (2000) exhibited considerable random variability about the fitted regression lines (R 2 values of 56.6% and 62.6% for Eqs. (5) and (6), respectively).…”

“…The use of correlations to estimate fish thickness and depth in lieu of measured dimensions is therefore likely to have introduced considerable variability into the model predictions, given that fish thickness and depth are the two shortest and therefore most critical dimensions for predicting cooling rates. However, assuming that the correlations of Bickers et al (2000) showed no systematic lack of fit to the dimensional data, use of these correlations in the chilling time prediction model would not be expected to result in a systematic overor under-prediction of chilling times. Both the slope and intercept of ln(Y) versus t predicted by the chilling model Flaked ice 29.7 ± 7.8 14.9 ± 6.4 6.9 ± 5.4 17.2 ± 4.1 Ice/water slurry 24.7 ± 6.3 16.2 ± 5.1 11.9 ± 4.8 17.6 ± 3.2 Combined 27.3 ± 5.0 15.5 ± 4.0 9.3 ± 3.6 17.4 ± 2.6 are dependent on fish thickness and depth dimensions (through R, b and Bi).…”

“…Both the slope and intercept of ln(Y) versus t predicted by the chilling model Flaked ice 29.7 ± 7.8 14.9 ± 6.4 6.9 ± 5.4 17.2 ± 4.1 Ice/water slurry 24.7 ± 6.3 16.2 ± 5.1 11.9 ± 4.8 17.6 ± 3.2 Combined 27.3 ± 5.0 15.5 ± 4.0 9.3 ± 3.6 17.4 ± 2.6 are dependent on fish thickness and depth dimensions (through R, b and Bi). Use of the correlations of Bickers et al (2000) is therefore expected to have contributed to the random error between predicted and measured slopes and intercepts, but should not have introduced significant systematic error.…”