battery model

[jack11]

I have a computer simulation of a RE system, which I use to size and validate a system before it's built.

I use a simple battery model in it, treat the battery as a huge capacitor C plus internal resistance R, and calculate the charge Q accumulation (and SOC) using 1st-order linear differential/integral equations (Q = time integral of current flowing in/out).

Then, instead of using Q and C to calculate the battery voltage V, I do Q-to-V mapping using a nonlinear function I picked up from some ieee proceedings paper.

Then, I calculate the energy stored in the battery using Q and V.

I have a feeling that treating a battery as a capacitor is not accurate, and that some fully-nonlinear model needs to be used to make the simulation more realistic.

Has anyone worked with battery models before, and can suggest something more realistic, but still easily implemented in a PC simulation?

[dnix71]

Not likely. You have to account for the type of battery (chemistry and wet/dry), internal temp, internal resistance, discharge curve for that type of battery (the non-linear IEEE curve you probably have), age of the battery and cabling and interconnects.

You also have to account for the type of series/parallel arrangement you have on the bank. You will always have an out-of-balance condition when discharging a series string because of differences between "identical" batteries. You also have consider that the rate and depth of discharge can greatly affect the useful life of some batteries.

Treating a battery as a capacitor isn't real, but reality can't be accurately modelled either.

[jack11]

thanks dnix,

I have several papers that delve into the issues you described, and more, and the modeling of these effects seems very complicated and often unclear to me. Also, I am not sure I'd gain a lot of improvement in accuracy by expending a lot of extra effort implementing these models.

The simulation gives results that more or less agree with the reality, so I guess for now I'll stick with what I have.