For the 24-h/48-h sampling strategy, this relative error could be calculated for any estimated EF, based on a number(n) of 24 h/48 h periods. These n 24-h/48-h periods were biosystems engineering 140 (2015) 79 e9 0 83 For each of the simulated sampling strategies and for each
value of n, 1000 simulations were executed, yielding a distri-
bution of the relative errors as a function of the number of sampling periods. Subsequently, the average (m) and the standard deviation (s) of this distribution were calculated for
each value of n. 95% confidence interval around the mean for ε are given by m 2s and m þ 2s. The confidence intervals of the relative error are a measure of the precision of the simulated
sampling strategy, while the average of the relative error represents the accuracy of the estimate.