|
发表于 2003-8-15 09:19:17
|
显示全部楼层
为什么我做的自然对流加有辐射模式和不加辐射模式,温度场算出的结果是一样的
Question: - How to accelerate the convergence of T3 for IMMERSOL radiation model ?
Answer:
Convergence of the energy (TEM1 or H1) and T3 equations can be strongly influenced by the PIL real variable RSG41.
RSG41 is a multiplier to the coefficient which is used to link the radiation temperature T3 to TEM1 in solids. The default value is unity, but this can be overwritten by setting RSG41>0 in the Q1 file.
A value of RSG41 equal to unity means that the term linking T3 and TEM1 in the solid is comparable with the surrounding conduction terms in the solid.
A value of RSG41>1 makes a stronger link between T3 and TEM1 than that of the surrounding conductive links. Thus sufficiently large values of RSG41 will achieve the desired effect of fixing T3 to TEM1. However, if the radiative contribution is large, slow and/or difficult convergence can be encountered.
Experience has shown that although smaller values of RSG41 promote convergence, they can artifically reduce the effects of radiation. This is because T3 and TEM1 will not be linked tightly together, and so most of the radiation will remain in the T3 equation and not be transferred to the TEM1 equation via the linking term.
The advice is to use a value of RSG1 in the range 1 to 100, or more generally as large a value as can be used to reasonably link T3 to TEM1 in the solid without preventing convergent behaviour.
In PHOENICS V3.4 onwards RSG41 is no longer required because a Partial-Elimination Algorithm (PEA) has been introduced whereby changes to temperature as a result of radiation fluxes are anticipated by the IMMERSOL model. The benefit is that the T3 and TEM1 values in solids are linked more efficiently, avoiding the conflict between accuracy and speed of convergence in earlier implementations of IMMERSOL. Experience with the PEA has shown that large temperature fluctuations can sometimes occur during early sweeps. There are various steps you can take to control this effect: (a) switch ISOLX, ISOLY and ISOLZ to zero (under 'Numerics', 'Iteration control'), (b) introduce VARMIN and VARMAX values for TEM1 and T3, (c) use FALSDT for TEM1 and T3, rather than LINRLX.
|
|