
Modelling the distribution process 127
then numerically solve equation 5.27 for H(t). We do that by computing
the right hand side at the time points 0.25, 0.5, 1 and then every hour up
to 12. Based on this data we now make a numerical deconvolution, using
the trapeze method, which provides us with estimates of H(t). The result is
shown in Figure 5.7 which also shows the analytical function for comparison.
FIGURE 5.7: True and numerically estimated integrated transfer function for
three-compartment model
To estimate I(h) is simple. It is given by H(0) = 22.4, and by numerical
integration of H(t) (again using the trapeze formula) we can estimate E(h)= ...