Production of surimi represents a potential utilization of a number of low-valued fish species, one of which, Pacific whiting, represents the largest biomass off the West Coast of the United States. However, a protease enzyme softens the fish flesh in Pacific whiting and limits the expansion of surimi production. Many studies have demonstrated the importance of time and temperature in minimizing the texture softening. An optimal design of the surimi seafood process is possible only when an accurate prediction of the time-varying temperature distribution throughout the surimi product can be obtained. This provides a measure of the heating rate and the extent of thermal processing. Such a prediction necessitates a study of the surface heat transfer coefficient which is one of the most important parameters for the heat transfer analysis. Associated with automated-machinery processing of surimi seafoods, a full understanding of the heat transfer coefficient (h) is especially important because high-quality surimi products using Pacific whiting only can be obtained through rapid and controlled heating. This study was intended to determine transient surface heat transfer coefficients in a steam heating environment, simulating the widely-used steam heating of thin-sheet surimi paste in the seafood industry. In determining the heat transfer coefficient, many different methods have been used including the inverse calculation method, the lumped mass method and the heat flux method. This study employed all three to measure and model the heat transfer coefficient (h) under similar steam conditions. A comparative evaluation was made to define the best method and model for the h determination. The inverse calculation method produced an h model which, when applied to a heat transfer analysis, provided the best agreement between predicted and experimental temperature profiles at three locations in surimi paste during a 1000-sec cooking period. The lumped mass method overestimated the heat transfer coefficients to food; the heat flux method gave inconsistent measurements. It is a classic inverse problem to estimate surface heat transfer coefficients from temperature measurements inside a product, a procedure which involves solution of the inverse heat conduction problem and parameter optimization. A whole domain function specification procedure was developed for the inverse calculation method. This procedure simulates heat transfer coefficients as specified functions by estimating all the unknown parameters in the functions over the total time interval. A nonlinear regression computer program was written for the inverse calculation of surface heat transfer coefficients, incorporating the implict Crank-Nicolson scheme for the finite-difference formulation of the one-dimensional heat conduction problem and the downhill simplex method for parameter optimization. This inverse calculation method provided relatively accurate models of the surface heat transfer coefficient. / Graduation date: 1997

Graduation date: 1997