To obtain solutions to these equations, we must specify boundary conditions. In this section, we define the commonly occurring boundary conditions.
B asic concepts
In a previous case study, we identified that the mass fluxes across the face of a membrane was equal on both sides. In the case study, we represented the concentration as a continuous function, which is an over-simplification.
More generally, when entering a new phase, there is a discontinuous shift in the concentration. We typically represent this using an equilibrium constant like so:
𝑐𝐴,1 (𝑥 = 0− ) = 𝐾𝑒𝑞 . 𝑐𝐴,2 (𝑥 = 0+ )
We may also identify a “bulk” concentration, that is, we know that near 𝑥 = 0, due to the transfer between the two phases, the concentration is varying there. However, after a relatively short distance away from 𝑥 = 0, the fluid may be well-enough mixed that we can identify a fixed value 𝑐𝐴1,𝑏𝑢𝑙𝑘 . It may be that this bulk concentration is changing over time; however, regardless of the time, when we are far enough from a source of disturbance to the concentration, it at least does not change in 𝑥.
1 | Page The film
If we have accepted that the concentration near the interface is varying only within this zone, we can propose a useful approximation. We let 𝛿 be the distance from 𝑥 at which we have reached 99% of the bulk concentration 𝑐𝐴,𝑏 , and call the fluid in the range 𝑥 ∈ [0, 𝛿] the “Film”. This is an imaginary region where most of the difference in the concentration is located.
Given that we have this nice easy concept of a film, we would like to write a nice simple expression like this:
𝑁𝐴,2 = 𝑘𝑐 (𝑐𝐴2,0 − 𝑐𝐴2,𝑏 )
What we actually still have is
𝑑𝑐𝐴 𝑁𝐴,2 = −𝐷𝐴𝑒 . [ ] 𝑑𝑥 𝑥=0+
However, since the curve nice and flat in this region, and given that we had defined the distance here to be 𝛿 we can write
𝐷𝐴𝑒 𝑘𝑐 = 𝛿 This now defines our mass transfer coefficient, and we see its direct connection with diffusivity and the film thickness. Correlations for 𝑘𝑐 abound in the literature.
2 | Page Note that we can determine the value of 𝛿 by solving the general flux expression:
⃗⃗⃗⃗ 𝑁𝐴 = −𝑐𝐷𝐴𝐵 ⃗∇𝑥𝐴 + 𝑥𝐴 (𝑁 ⃗⃗⃗⃗𝐴 + ⃗⃗⃗⃗⃗ 𝑁𝐵 ) For example, in the earlier example concerning the diffusion of component 𝐴 through stagnant 𝐵 which we showed had the ffg solution
𝑐𝐷𝐴𝐵 1 − 𝑦𝐴 𝑁𝐴 = ln 𝑧 − 𝑧0 1 − 𝑦𝐴0 Or 𝑁𝐴 (𝑥 − 𝑥0 ) 𝑦𝐴 = 1 − (1 − 𝑦𝐴0 ) exp 𝑐𝐷𝐴𝐵 We note here that if we happen to know 𝑦𝐴 and 𝑦𝐴0 , we can solve for the distance (𝑥 − 𝑥0 ).
In this case 𝑧 − 𝑧0 = 𝛿 and 𝑦𝐴 = 0.99 𝑦𝐴,𝑏𝑢𝑙𝑘 . We can therefore solve for 𝛿 and hence relate 𝑘𝑐 to 𝐷𝑒 . In practice, 𝑘𝐶 is measured or estimated by correlation. The point is that we may then write the boundary condition
Cap 6 (Student Mathematical Library 94) James Bisgard - Analysis and Linear Algebra_ The Singular Value Decomposition and Applications-American Mathematical Society (2021)
