[3.1.1.1] The approach presented here is based on the traditional theory presented in Section 1 combined with the distinction between percolating and nonpercolating phases introduced in Refs. [14, 16, 15] and discussed in Section 2. [3.1.1.2] The volume fractions of the subsets and are denoted as . [3.1.1.3] Let denote the porosity (volume fraction of ). [3.1.1.4] Volume conservation implies the relations

(16a) | |||

(16b) | |||

(16c) |

where are volume fractions, and are saturations. [3.1.1.5] The water saturation is defined as and the oil saturation as .

[3.1.3.1] As in the traditional theory the porous medium is again assumed to be macroscopically homogeneous so that eq. (4) holds. [3.1.3.2] The flows are slow, and hence also eq. (7) continues to hold without change. [3.1.3.3] For incompressible fluids one has now

(17a) | |||

(17b) | |||

(17c) | |||

(17d) |

analogous to eq. (5a).

[3.1.4.1] The momentum transfer into phase from all the other phases is assumed to be a simple viscous drag,

(18) |

where the resistance coefficient quantifies the viscous coupling between phase and . [3.1.4.2] For the rigid rock matrix [page 4, §0] . [4.0.0.1] Hence is the momentum transfer from the wall into phase . [4.0.0.2] Then

(19a) | |||

(19b) | |||

(19c) | |||

(19d) |

where and was used because there is no common interface and hence no direct viscous interaction between these phase pairs. [4.0.0.3] Each is a -matrix.

[4.0.1.1] In the present approach the body forces are given by gravity plus capillary forces

(20a) | |||

(20b) | |||

(20c) | |||

(20d) |

contrary to eqs. (8). [4.0.1.2] The capillary body forces are responsible for keeping the trapped fluids inside the medium. [4.0.1.3] They are obtained as gradients of capillary potentials

(21a) | |||

(21b) |

where the capillary potentials are defined as

(22a) | |||

(22b) |

with constants and exponents .

[4.0.2.1] Next the stress tensor for percolating phases can be specified in analogy with eq. (6) as

(23a) | |||

(23b) |

where and are the fluid pressures. [4.0.2.2] The stress tensor for the nonpercolating phases cannot be specified in the same way because the forces cannot propagate in nonpercolating phases. [4.0.2.3] Here it is assumed that these stresses are given by the pressure in the surrounding percolating phase modified by the energy density stored in the common interface with the surrounding percolating phases. [4.0.2.4] This suggests an Ansatz [4]

(24a) | |||

(24b) |

where is the oil-water interfacial tension, and the unknowns are the interfacial areas per unit volume of porous medium between phases and , resp. and . [4.1.0.1] (The factor arises from the definition of per unit volume of porous medium.) [4.1.0.2] To avoid equations of motion for the unknowns and it is assumed that geometrical relations of the form

(25a) | |||

(25b) |

hold, where are prefactors assumed to be constant. [4.1.0.3] Thus for the nonpercolating phases

(26a) | |||

(26b) |

will be used below. [4.1.0.4] Here

(27a) | |||

(27b) |

are constants.

[4.1.1.1] The mass transfer rates must depend on rates of saturation change. [4.1.1.2] They are here assumed to be

(28a) | |||

(28b) |

where are constants. [4.1.1.3] The parameters , are defined by

(29a) | |||

(29b) | |||

(29c) | |||

(29d) |

[page 5, §0] where are limiting saturations for and are equilibration time scales for reaching capillary equilibrium. [5.0.0.1] For simplicity

(30) |

will be assumed below.