Energy balance in a pistoncylinder system
Hi, I am simulating the behaviour of oscillating flow and heat transfer in a pistoncylinder system using CFX11.0. The 2D axisymmetric model includes a moving boundary and isothermal boundary conditions are applied. My problem is that the energy balance of a whole cycle can't be reached while temperature and pressure seem to get to a steady state. In a simple pistoncylinder system, the net PV work during a cycle should be equal to the neat heat in and out of the system as the system gets to a steady state. In my cases, the net heat is nearly 2~9 times larger than the net PV work. It is noted that the net PV work is calculated by integrating PdV over a cycle and the net heat is calculated by integrating the heat flux over whole boundary area using CFXpost function and time period. I had simulated the MIT single space experiments using CFX. The results of pressure agree well with the experimental data, but the net loss is a few times larger than the net PV work over a cycle.
Any help will be highly appreciated. Best regards Taotao 
Re: Energy balance in a pistoncylinder system
How many cycles you have done in your calculation? If energy balance is not reached, net heat flux number is bogus. Is mass flow balanced? Temperature reaches stable condition while energy is not balanced. Things do not add up.

Re: Energy balance in a pistoncylinder system
1. Are you solving for "Total Energy" or "Thermal Energy"? 2. Have you converged within each timestep? 3. How many coefficient loops is the solver using? Does it reach the limit of coefficient loops you have provided? 4. What are you calculating in Post?

Re: Energy balance in a pistoncylinder system
Hi Longbow
Thank you for your response. I had calculated 120 cycles. I don't know how to check mass flow balance because this is a closed system. Regards Taotao 
Re: Energy balance in a pistoncylinder system
Hi Cyclone,
Thank you for your response. 1. I am solving for Total energy. 2. The solution is converged under RMS 1e6 3. The number of coefficient loops is 10 and it converged less than the limit. 4. I am using the function of AreaInv(heat flux)@boundary and then integrate it over the whole cycle to calculate the net heat loss. Regards Tonny 
Re: Energy balance in a pistoncylinder system
Hi,
Are you using a moving mesh with second order time differencing? There has been some problems with heat conservation with fully compressible fluids in moving mesh simulations with second order timestepping in previous versions of CFX, not sure if the current version still has the problem. Glenn Horrocks 
Re: Energy balance in a pistoncylinder system
Hi Glen,
Thank you for your response. Yes, I am using a moving mesh with second order time difference. I am looking at heat transfer in a pistoncylinder system by using CFX11.0. Any suggestion for dealing with this problem? Regards Tonny 
Re: Energy balance in a pistoncylinder system
Hi,
Not recommended. Use first order time differencing. At least try it and see if it gives you better conservation. Glenn Horrocks 
Re: Energy balance in a pistoncylinder system
Hi Glenn,
Thank your very much for your help. I am going to try it using first order time difference. Why is CFX not recommended for this kind of problem? Is it mainly due to moving mesh? Which commercial package such as Fluent, Star CD is more suitable for this kind of problem? Tonny 
Re: Energy balance in a pistoncylinder system
Hi,
No, the "not recommended" refers to the combination of second order time stepping, moving mesh and compressible flow. The easy fix is to go to first order time stepping. CFX can model this type of flow just fine, as long as you are careful and do the normal checks (which you have to do with any software anyway). Glenn Horrocks 
Re: Energy balance in a pistoncylinder system
You don't have inlet and outlet in your model? In other words, you have a completely closed cylinder with piston compressing and decompressing fluid inside? Is piston isothermal boundary too?

Re: Energy balance in a pistoncylinder system
I forgot about that limitation. This has been fixed for version 12, so second order time discretization can be used.

Re: Energy balance in a pistoncylinder system
Hi Longbow,
No, I don't have inlet and outlet boundary. The piston is isothermal boundary too. Thanks Tonny 
Re: Energy balance in a pistoncylinder system
Hi Glenn,
I calculated it using firstorder time differencing scheme which gave better results than secondorder time differencing scheme, but the results doesn't seem to be satisfying while you look at heat transfer in the system. Tonny 
Re: Energy balance in a pistoncylinder system
Hi Cyclone,
I tried version 12.0 (beta) with second order time difference scheme. The results for energy conservation is similar to those from version 11.0 and the convergence is even worse. Tonny 
Re: Energy balance in a pistoncylinder system
Hi,
It was found the area of side wall is constant while I extracted the area of side wall from CFXpost using the internal function of area()@, but the volume of domain varies with time.It doesn't make sense. Have you noticed that problem? Many thanks Tonny 
Re: Energy balance in a pistoncylinder system
Hi,
What do you mean "results don't seem to be satisfying"? You mean you have a heat imbalance? Have you done a convergence sensitivity check? Have you done a timestep sensitivity check? I bet you have too big timesteps and/or incomplete convergence. Glenn Horrocks 
Re: Energy balance in a pistoncylinder system
Hi Glenn,
Sorry for slow response. I have been doing timestep sensitivity check recently using first order time differencing. I found the net heat flow of one cycle strongly depend on the timestep. Up to now, I still have a heat imbalance of 30%. In addition, it make convergence more difficult using first order time differencing. It was found the area of side wall is constant while I extracted the area of side wall from CFXpost using the internal function of area()@, but the volume of domain varies with time.It doesn't make sense. Have you noticed that problem? Thank you very much for your help Taotao 
All times are GMT 4. The time now is 00:16. 