Discussion Forum

Hi

I would answer: pls be more specific, sensitivity to what ?

isnt that basically the value of the elasticity or compliance tensor ?

--
Good luck
Ivar

Hello Ivar,

Thank a lot for replying. Voltage sensitivity is expressed as Voltage generated in the PZT for every 'g' input acceleration. The unit for voltage sensitivity is Volts/ g, where g is 9.8m/s^2. Charge sensitivity is the charge generated in the PZT for every 'g' of input acceleration. The unit for charge sensitivity is pF/ g. Here pF is pico Farad.

It would be of immense help if you could suggest me a way to simulate these parameters.

Thanks a lot.

-Sankha

Hi

so that is for a PZT used as an accelerometer with an inertial mass and a base acceleration ;)

And I assume then your two values are linked via the PZT capacitance, and or any stray capacitance, so for a given geometry and material parameters, if you have one, you have the other

I would then start with a simple PZT squeezed between two small masses with a linear acceleration (in 2D) and compare the results withan existing accelerometer.

Then I would start 3D and to think how to modify the model geoemtry and materials, and test different combinations mass/pzt in compression, shear etc, and see the sensitivities for different acceleration directions (cross coupling etc)

Nice project

--
Good luck
Ivar

Hello Ivar,

Thanks for replying. You are absolutely correct. I am trying to simulate voltage sensitivity of an accelerometer. The problem is that I do not know the steps to get these numbers. I will tell you what I have done till date, and you can tell me whats left.

I have used piezodevice module and have done Eigen frequency analysis. So I know the natural frequencies. Since it is voltage/g, I think that Eigen frequency analysis will not work. Should I use Frequency domain? Do I have to integrate the voltage field over all the domains? I see people reporting these values in journals but they do not discuss the methodology adopted in FEM analysis!!!!

It will be very kind of you to tell me the steps to be followed to do this analysis.

-Sankha

Hi

If you have a representative geoemtrical model, with correct material data, particularly for the PZT, and perhaps if you include the wire + sensor input impedance/load, then you should indeed be able to use a frequency domain scan to see how the systems responds. Normally, one uses an accelerometer in the DC to 1/2 of the first eigenfrequency region, it becomes too non linear when approaching the resonance. In anycase an eigenfrequency analysis is essential to define the first resonance.

With the frequency domain sweep you can define your input load and it should work. Note you cannot (I have not managed so far) to use the frequency domain - Modal with prescribed acceleration inputs, only with forces, so I would suggest to stick first to a standard frequency domain scan

--
Good luck
Ivar

Hello Ivar,

Thank you for your reply. However, I have few confusions mentioned below,

1) What is "wire + sensor input impedance/load". How do I implement these elements in Comsol. I have a geometric model of the accelerometer but I have not implemented these yet.

2) What sort of load should I use for the frequency domain sweep? Do you mean acceleration by load? Then I can start from 1 g and go upto 10 g.

3) What are the post processing steps involved for obtaining these results.

Do you have any example model which you can share with all of us. This would help us a lot in understanding the simulation steps.

Once again many thanks for providing your time to help me.

-Sankha

Sankha,

I was following your conversation with Ivar and thought I'd add some comments + example as you requested. Although if you already have a model built, it would be much easier to work directly on your file.

What I attached shows how to read the voltage from a floating potential when 1g (9.81 m/s2) of acceleration is applied to the base. I don't know if you'll be able to get the charge sensitivity off this model as when you set up the floating potential, it directly asks you for the charge...which is a constant (I set =0), so that makes me think you'll need a different model to find charge sensitivity. I'm not sure what exactly that would be...you could get capacitance as a function of frequency from changing the floating potential to a 1V terminal and taking imag(pzd.Y11)/(2*pi*freq) ....although that would not be per G, but per Volt. Maybe (ok, this is a guess), you can find it from the combination of these two parameters, i.e. C/V * V/g ? Or set up a circuit (cir physics) with the cables, etc. as Ivar suggested?

Dear Marc,

Thanks a lot for sharing your model. Is this model built in Comsol 4.2 or it's later versions? The Comsol license I have is 4.1!!!!!!!!!! I could not open your model.

If possible, can you kindly convert your file compatible to Comsol 4.1?

Thank you again for the help.

-Sankha

Hello Marc,

I have been able to obtain the voltage sensitivity of the accelerometer using your technique. However, as you had mentioned I do not obtain any charge (I end up with a o charge accross the frequency spectrum) development.

Can you describe your thoughts about this? Can you give some ides as to how do I go about creating the circuits to get the charge sensitivity, capacitance!!

-Sankha

Hello Ivar,

I am working on this problem for quite some time. Although I have achieved some success in obtaining Eigen frequencis, Frequency response, Voltage sensitivity, I am still to be able to simulate the charge accumulation of a piezoelectric accelerometer as a function of acceleration. I have searched through out this blog, Comsol website but unfortunately there is not a single solved example to help me with this. It would be of real help if you could upload/share an example for simulating charge sensitivity of piezoelectric accelerometer [or energy harvester].

Thanks,

Sankha

Sankha,

There is probably a better way to do this, but you may want to try coupling your piezo to a simple electrical circuit (cir) physics by setting up a piezo terminal boundary to "Circuit" type (I don't think you can automatically couple to a floating potential BC), then adding a voltage source with 0 Volts input. If the solver doesn't give you a singular matrix error (it might) you should be able to read the current off the dummy voltage source.

-Marc

Hello Marc and Ivar,

Thank you both for all the cooperation you have provided in last couple of weeks. I have built 2 models (one for voltage sensitivity and one for charge sensitivity) of a cantilevered piezoelectric beam. To obtain charge development I used electrical circuit physics and connected the terminal of piezoelectric physics with a capacitor. I am attaching these two models. It will be kind of you to look at these two models and let me know if there are mistakes.


Thank you,

Sankha

Hi

looks nice, a few suggestions:

1) you have the COMSOL build in constant "g_const" = 9.81 m/s^2, see help search "physical constants" (do not forget to add the suffix "_const" to all the variable symbol names given in the table

2) to better resolve your plot position where you measure use the Marker Cycle, Positioning In data points for the Frequency domain sweep

3) as you are using a terminal BC, your surface will be at the same potential so there is no nead to use a multi point graph. In fact its easier to access these values asa Derived Varaible Global Evaluation port and terminal outputs adn tehn plot the charge, current and voltage, but as these might be complex, plot them in amplitude abs() and phase atan2(imag(),real()) mode separately

4) most imporant, if you add the "CIR" link to your Terminal BC condition you need to add a CIR physics too, at least with a resistace and a I vs U link to the Terminal, this applies for both casess, but your CIR is only missing the the voltage model, when I load it

5) I would use rather a smaller frequency sweep something like 10^{range(2,0.01,3)}

--
Good luck
Ivar