Regression models

Fixed-term models using deadweight and storage capacity provide a robust model fit:

         Storage new = 0.0041 DWT + 425, R2 = 0.66

         Storage conv = 0.0055 DWT + 115, R2 = 0.77.

Storage is described in Mbbl and DWT is expressed in tonnes. The poorer fit model for purpose-built hulls is attributed in part to the greater diversity of newbuild designs.

Models without an intercept are as follows:

          Storage new = 0.0058 DWT, R2 = 0.44

          Storage conv = 0.006 DWT, R2 = 0.76.

Circular FPSOs were not included in the sample, and if gas FPSOs are excluded, the model fits improve since the samples are more homogenous.

Deadweight correlates with storage capacity, but one should always be mindful that when comparing individual units, a smaller deadweight FPSO could have greater storage capacity than a heavier deadweight unit.

Storage vs dimensions

Hull size determines cargo capacity, and the greater the dimensions of the hull the greater the storage capacity.

Storage capacity is correlated to the hull’s length L, hull effective area L*W, and hull block volume L*W*D as follows:

       Storage = 7.63 L – 949, R2 = 0.50

       Storage = 0.101 L*W – 244, R2 = 0.62

       Storage = 0.0028 L*W*D + 145, R2 = 0.63.

       Storage is described in Mbbl and all dimensions are measured in meters.

Like the previous deadweight relations examined in Part 3 of this FPSO series, model fits improve with each additional variable because each new variable is both relevant and (mostly) independent from the previous variables:

      Storage = 4.43 L, R2 = 0.89

      Storage = 0.085 L*W, R2 = 0.93

      Storage = 0.003 L*W*D, R2 = 0.93.