I need to ask a question about partial derivatives. I want to solve this equation (steady state, one dimensional continuity equation): $$\frac{\partial (\rho u)}{\partial z}=0$$ which is equivalent to: $$\rho\frac{\partial u}{\partial z}+u\frac{\partial \rho}{\partial z}=0$$ and discretized by means of finite differences: $$\rho_i\frac{(u_i-u_{i-1})}{\delta z}+u_i\frac{(\rho_i-\rho_{i-1})}{\delta z}=0$$ along with initial conditions like: $$z=0 \quad \rho_i=\rho_0 \quad , \quad u_i=u_0$$

My questions are: Does this equation has one unique solution? I mean one could increase $u$ and decrease $\rho$ (or the other way around) and have infinite solutions right?
In this case this equation can describe the conservation of the quantity $(\rho u)$. What about the conservation or $u$ or $\rho$?
If I could calculate the $\rho$ (density) variable through lets say an equation of state (like $PV=nRT$) which would be variable in $z$ direction can i rewrite the first equation like: $$\rho\frac{\partial u}{\partial z}=0$$ $$\rho_i\frac{(u_i-u_{i-1})}{\delta z}=0$$ which do have one unique solution.

Just to let you know of my intention, I eventually want to solve a set of PDEs: $$\frac{\partial \rho}{\partial t} +\frac{\partial (\rho u)}{\partial z}=0$$ $$\frac{\partial C_i}{\partial t} +\frac{\partial (u C_i)}{\partial z}=Reactionrate$$ $$\rho=MW_{mean}\frac{P}{RgT}, MW_{mean}=\sum^{N_{comp}}_{i=1}(Y_i*mw_i), Y_i=\frac{C_i}{\sum^{N_{comp}}_{i=1}{C_i}}$$

  • 4
    $\begingroup$ In general, it's better to apply the finite difference scheme without using the product rule. You will find that the resulting method is more stable and preserves continuity. $\endgroup$
    – Paul
    Jan 26, 2016 at 18:50
  • $\begingroup$ Paul, thank you for your reply. I have read about the difference of conservative and non conservative form see link. The think is that if I treat the quantity ρ*u like one variable then in the case of dynamic balance I would have another one (ρ). Is n't that correct? I would need more equations than the ones that i have now. Maybe one to correlate ρ & u? $\endgroup$
    – ASK22
    Jan 26, 2016 at 21:13
  • $\begingroup$ Also, please refer to this link for a more detail explanation of my problem. $\endgroup$
    – ASK22
    Jan 26, 2016 at 21:30
  • $\begingroup$ I think you should put your follow-up questions into a question of itself. $\endgroup$ Feb 3, 2016 at 15:20
  • $\begingroup$ @WolfgangBangerth. You mean like a whole new question? or like an answer to this question? $\endgroup$
    – ASK22
    Feb 3, 2016 at 15:21

1 Answer 1


The equation $\frac{\partial}{\partial z}(\rho u)=0$ only has a unique solution for the product $\rho u$ (namely a constant), but not for each factor separately. You need a different equation to tease them apart.

In your context, the equation describes the conservation of momentum. You will need a separate equation for the conservation of mass, which is going to have a term $\frac{\partial}{\partial z}\rho$. Together, these two equations are going to uniquely define both velocity and density.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.