Guignon B, Torrecilla J S, Otero L, Ramos A M, Molina-García A D, Sanz P D
Department of Engineering, Instituto del Frío (CSIC), MALTA CONSOLIDER TEAM, c/José Antonio Novais, 10. 28040 Madrid, Spain.
Crit Rev Food Sci Nutr. 2008 Apr;48(4):328-40. doi: 10.1080/10408390701347736.
The Pure water (P,T)-phase diagram is known in the form of empirical equations or tables from nearly a century as a result of Bridgman's work. However, few data are available on other aqueous systems probably due to the difficulty of high-pressure measurements. As an alternative, six approaches are presented here to obtain the food phase diagrams in the range of pressure 0.1-210 MPa. Both empirical and theoretical methods are described including the use of an artificial neural network (ANN). Experimental freezing points obtained at the laboratory of the authors and from literature are statistically compared to the calculated ones. About 400 independent freezing data points of aqueous solutions, gels, and foods are analysed. A polynomial equation is the most accurate and simple method to describe the entire melting curve. The ANN is the most versatile model, as only one model allows the calculation of the initial freezing point of all the aqueous systems considered. Robinson and Stokes' equation is successfully extended to the high pressures domain with an average prediction error of 0.4 degrees C. The choice of one approach over the others depends mainly on the availability of experimental data, the accuracy required and the intended use for the calculated data.
由于布里奇曼的工作,纯水的(P,T)相图以经验方程或表格的形式为人所知已近一个世纪。然而,其他水体系的数据可能因高压测量的困难而很少。作为替代方法,本文提出了六种方法来获得压力范围为0.1 - 210 MPa的食品相图。文中描述了经验方法和理论方法,包括人工神经网络(ANN)的应用。将作者实验室获得的实验冰点以及文献中的实验冰点与计算得到的冰点进行了统计比较。分析了约400个水溶液、凝胶和食品的独立冰点数据点。多项式方程是描述整个熔化曲线最准确且简单的方法。人工神经网络是最通用的模型,因为只需一个模型就能计算所考虑的所有水体系的初始冰点。罗宾逊和斯托克斯方程成功扩展到了高压领域,平均预测误差为0.4摄氏度。选择一种方法而非其他方法主要取决于实验数据的可用性、所需的精度以及计算数据的预期用途。