Quantcast
Channel: Latest Discussions - COMSOL Forums
Viewing all articles
Browse latest Browse all 26527

Failed to Evaluate Jacobian Variable

$
0
0
Dear COMSOL users,

I am trying to model the electromagnetic heating stimulation for heavy oil reservoir (2D radial) to be further studied for my bachelor's degree thesis.

The model is very simple. I am coupling "Heat Transfer" and "Darcy's Law" to draw two desired plots: pressure vs time vs radius and temperature vs time vs radius.

1. VISCOSITY

Ideally, viscosity decreases when temperature increases (which is the core philosophy of electromagnetic heating). Therefore I used the below function for viscosity in the Darcy Law model:

viscosity = teta4 = 4*10^(7)*exp(-0.053*T)

When I ran the calculation, the following error message came up:

"The following feature has encountered a problem:
Feature: Time 1 (sol1/t1)
Error: failed to evaluate variable Jacobian."

Surprisingly, when I curiously replaced the viscosity in the Darcy Law model with this random function:

viscosity = 3.78 + T/10000,

the calculation by the Solver Sequence worked smoothly. My further experiment has stipulated that the Solver Sequence will fail to evaluate variable Jacobian if the viscosity decreases as temperature increases, while it will work smoothly if the viscosity increases as temperature increases (which is not logically fulfilling).

2. HEAT SOURCE

The same error message (failure to evaluate Jacobian variable) came up when I input this function into the Qb for boundary heat source in the Heat Transfer model:

Qb = qem = alpha*power*exp(-alpha*(r-rw))/r,

while power = constant, rw = constant, and alpha = 2*pi*freqtrick*(epsilon*mperm/2)^0.5*((1+(sigma/(epsilon*2*pi*freqtrick))^2)^0.5-1)^0.5[1/m],

while freqtrick = constant, epsilon = constant, mperm = constant, and sigma = 10^(-2)*(1+26.04*10^(-3)*(T-297.15)+(13.12*10^(-3)*(T-297.15))^2+((-9.97)*10^(-3)*(T-297.15))^3). In short, qem is temperature-dependent because sigma is also temperature-dependent.

Surprisingly, again, when I curiously replaced the viscosity in the Darcy Law model with this random function:

viscosity = 1000-T/1000,

the calculation by the Solver Sequence worked smoothly. This time, unlike the case of viscosity above, I cannot really determine the pattern.

QUESTION

What should I do to enable my model works with these functions:

1. Darcy's Law: viscosity = teta4 = 4*10^(7)*exp(-0.053*T)

2. Heat Transfer: Qb = qem = alpha*power*exp(-alpha*(r-rw))/r, provided that

alpha = 2*pi*freqtrick*(epsilon*mperm/2)^0.5*((1+(sigma/(epsilon*2*pi*freqtrick))^2)^0.5-1)^0.5[1/m], provided that

sigma24*(1+26.04*10^(-3)*(T-297.15)+(13.12*10^(-3)*(T-297.15))^2+((-9.97)*10^(-3)*(T-297.15))^3).

I cannot attach my model because the file size is too big. Thank you for your help.

Regards,
Yosaka Eka Putranta


Viewing all articles
Browse latest Browse all 26527

Trending Articles