Meaningful energy analysis of industrial processes requires the installation of energy monitoring systems. However, a lack of systematic methods for identifying the required measurement points, joint to scarce information on the related benefits, results in reluctance from companies in investing in such systems.

The paper presents a method for identifying necessary measurement points in industrial processes. It is applied on four milk powder production plants. It combines Pinch Analysis with uncertainty analysis (Monte Carlo simulations), global sensitivity analysis (standardised regression coefficients) and optimisation procedures to solve the “Factor Fixing” and “Variance Cutting” problems. In this way it identifies a limited number of parameters whose precise and accurate measurement is paramount to meaningfully characterise the plant from an energy perspective.

Comparing the results obtained from the four case studies it was possible to infer some general traits of milk powder production processes. In particular, 15 out of 60 to 66 parameters were identified as generally important, their position in the plant was highlighted and the minimum accuracy level required for their measurement was estimated. Such information could subsequently be used for designing an energy monitoring system and giving proof of its benefits to the involved company, by quantifying them in terms of uncertainty reduction in the outcome of the energy analysis.

对工业过程进行有意义的能源分析需要安装能源监控系统。但是，由于缺乏用于识别所需测量点的系统方法，加上有关益处的稀缺信息，导致公司不愿投资此类系统。

本文提出了一种识别工业过程中必要的测量点的方法。它应用于四个奶粉生产厂。它将捏分析与不确定性分析（Monte Carlo模拟），全局灵敏度分析（标准化回归系数）和优化程序相结合，以解决“系数固定”和“方差切割”问题。通过这种方式，它可以识别数量有限的参数，这些参数的精确和精确测量对于从能源角度有意义地表征设备至关重要。

比较从这四个案例研究获得的结果，可以推断出奶粉生产过程的一些一般特征。特别是，在60到66个参数中，有15个被确定为普遍重要的参数，突出显示了它们在工厂中的位置，并估计了其测量所需的最低精度。此类信息随后可用于设计能源监控系统，并通过在能源分析结果的不确定性降低方面对其进行量化来证明所涉及公司的利益。