Previous research described the use of machine learning algorithms to predict aircraft fuel consumption. This technique, known as Virtual Sensors, models fuel consumption as a function of aircraft Flight Operations Quality Assurance (FOQA) data. FOQA data consist of a large number of measurements that are already recorded by many commercial airlines. The predictive model is used for anomaly detection in the fuel consumption history by noting when measured fuel consumption exceeds an expected value. This exceedance may indicate overconsumption of fuel, the source of which may be identified and corrected by the aircraft operator. This would reduce both fuel emissions and operational costs. This paper gives a brief overview of the modeling approach and describes efforts to validate and analyze the initial results of this project. We examine the typical error in modeling, and compare modeling accuracy against both complex and simplistic regression approaches. We also estimate a ranking of the importance of each FOQA variable used as input, and demonstrate that FOQA variables can reliably be used to identify different modes of fuel consumption, which may be useful in future work. Analysis indicates that fuel consumption is accurately predicted while remaining theoretically sensitive to sub-nominal pilot inputs and maintenance-related issues.