Latest Discussions - COMSOL Forums

To generate the Transient response curve( Time Versus Amplitude plot )

Thu, 02 May 2024 08:07:04 +0000

I am an amateur at COMSOL multiphysics. I am currently running simulations of SAW resonators.

I completed the SAW frequency domain analysis (with the help of SAW tutorials and the COMSOL library) and am now going on to the time domain analysis. More specifically, I employed the base surface of Lithium Tantalate and Lithium Niobate wafers, as well as Al and Gold IDT (Inter Digitised Electrodes), to generate Surface Acoustic Waves.

I am sharing two images: the IDT dimensions and the literature copied (Courtesy) plot.

My question is how I can acquire the time vs amplitude response from the comsol( TRANSIENT RESPONSES) , and which physics / study should I choose? In the time-dependent solution I tested with Sinwt as input and the result were negative.

heat transfer with surface to surface radiation and solid mechanics

Thu, 02 May 2024 07:34:13 +0000

Good morning, I have to simulate the thermal effect of a box bridge subjected to solar radiation. Selecting only the external edges as surfaces subject to irradiation and subject to a thermal flow, the program gives me the following error: "External sources are radiating on boundaries for which only the opposite side has well-defined material properties. This may be due to erroneous definition of the boundary radiation directions or the opacity of the adjacent domains. comp1.rad.Fextbackside can be plotted to show the boundaries that do not have well-defined material properties."

the drawing of the box in question is a section and therefore is not a closed section. Another question, how can I tell the program that I don't have air recirculation in the hollow part?

Decreasing Drag Coefficient (Cd) with Increasing Mesh Quality (3D sphere in a box, turbulent flow)

Wed, 01 May 2024 10:58:59 +0000

Hello COMSOL Community,

I'm working on a project where I need to calculate the drag coefficient (Cd) of a 3D sphere positioned inside a box. I have experimented with varying mesh qualities, starting from a coarse mesh (physics-controlled mesh) to assess the balance between accuracy and computational effort.

However, I'm encountering an issue where the calculated drag coefficient (Cd) consistently decreases as the mesh quality improves. For my calculations of the drag force, I am using "Id(?∗spf.nymesh)" where Id is the integration over the sphere’s surface to compute the drag force.

My drag coefficient Cd is calculated by: 2/(AUin^2rho0)Id(pspf.nymesh). A is set to pi*R^2.

As the mesh is refined, the value of Cd gets progressively smaller. Can anyone help me understand why the drag coefficient decreases with better mesh quality?

Any insights or suggestions on how to tackle this issue would be greatly appreciated.

Thank you!

Wish to change circular coil to hexagonal shape

Wed, 01 May 2024 09:07:25 +0000

Hello There,

I need a help to change my circular coil to hexagonal shape.Take a look the editing area here, I am adding my design file here.

Growing/Expanding domain

Wed, 01 May 2024 08:31:56 +0000

Hi,

I'm working on a three-layer model where my middle domain will grow as a function of time (see the blue boundary in image), I have a problem how to find or define a physics or a method to make this middle domain grow knowing that I use the temporal solver.

How to add a conditional displacement control of points?

Tue, 30 Apr 2024 13:14:51 +0000

Suppose there is a range of displacement, let's say [0,5]. What I want to do is that in this range the point can move freely and when it is exceeding the boundary, add a displacement control to force this point stay in this boudary. Any methods to implement this?

relative tolerance

Tue, 30 Apr 2024 13:08:19 +0000

What does relative tolerance mean in time dependent study setting?

Energy diagram of TiO2/Si heterostructure

Tue, 30 Apr 2024 11:49:55 +0000

I am a beginner in the semiconductor physics and would like some help with setting up simulation. I would like to simulate band diagram of thin film of TiO2 on Si wafer substrate. It is not a device, only thin film of anatase TiO2. So, can you suggest on how to get started

How to define indicator states in event interface using solution of previous step time?

Tue, 30 Apr 2024 04:51:25 +0000

I am trying to use event interface to re-initialize the boundary displacement when the strain energy approaches constant.

To fulfill this, I first define the strain energy Us=intop1(us), where 'intop' is a defined integration operator and 'us' is the strain energy density. Then, four sample points are defined as Usp1=withsol('current',Us,setind(t,-nprev)), Usp2=withsol('current',Us,setind(t,-2nprev)), Usp3=withsol('current',Us,setind(t,-3nprev)),Usp4=withsol('current',Us,setind(t,-4*nprev)), where nprev is set to 200. Hopefully, this four variables could sample the previous step time strain energy values.

Next, to monitor the variance of Us with moving time window, another variable is defined as Usvary=if(t>0,try_catch(abs(Usp1+Usp2+Usp3+Usp4-4*Us)/Us,1),1). All this variables are defined through 'component 1-->varaibles 1'.

In the Event Interface, an indicator variable is defined as Usflag=Usvary/Ustol-1. Ustol is a prescribed tolerance, within which the strain energy will be regarded as constant and the event will be trigged. The event condition is specified as 'Usflag<0'.

However, during the trial run, I find that Usvary is evaluated as 1 throughout the whole simulation. It seems something is wrong when I try to access the strain energy of previous step times.

I would appreciate it if anyone could give me some hints. Thank you so much!

Flow Uniformity in microfluidics channels using optimization function

Tue, 30 Apr 2024 00:18:48 +0000

Hi Everyone,

I am trying to use the parameter optimization function on comsol to optimize the width of four channels. The goal is to achieve uniform flow rate by varrying the resistance of the channels.

I added boundary probe for spf.u at an equal distance from the outlet and set these boudary probes as the objective functions. Since the objective function doesn't allow a contant value I set the variable for all the boundary probes as such (comp1.bnd1-1e-2)^2. Then I added the pre-defined width parameters (W1-W4) in the control variables and parameters section.

I ran the optimization function using Nelder-Mead method, with minimization function. The width of the channels marginally changed, but it did not achieve uniformity like I expected.

Does anyone know how to correctly set-up the objective function to have all the boundary probes have the exact same value?

Thank you

Piezoresistive cantilever beam

Mon, 29 Apr 2024 14:03:47 +0000

I would like to get the plot of resistance change vs strain plot from a cantilever beam made of piezoresistive material. I have fixed an end of it and am applying the load at the other end. I have also taken one end as the ground and other end of it as a terminal with 1V.

Am I missing something. I am facing an issue in the plot. Any help would be appreciated.

thank you

About The 'Reuse Solution From Previous Step' In Parametric Sweep And Time-Dependent Solver

Mon, 29 Apr 2024 12:29:41 +0000

Hi there,

I am trying to modeling the buckling and post-buckling process using the parameteric sweep and time-dependent solver like in this presentation about psudo-dynamic approach. We may take the buckling of a compressed bar as an extremely simplified example. Suppose a bar measured $L$ is fixed at the left end and loaded through displacement control in terms of nominal compressive strain $ee$ at the right end (as shown by the graph below). Due to the strong nonlinearity of the original problem, the steady problem is passed to a time-dependent one like in this presentation. The modeling is based on the time-dependent solver with parameteric sweep over $ee$. In each time-dependent study step, initial conditions like $w_0=0.1(1-cos(16{\pi}x/L)), u_0=-eex$ are prescribed and then the system is allowed to evolve for a sufficiently long time period $T$, say, $T=5000s$.

During the trail run, I encounter several questions:

1) The system doesn't seem to start from the initial conditions. For example, the result of $w$ at $ee=0, t=0s$ (shown belown) is quite different from the prescribed initial condition. And the $w$ profile at $t=0$ changes as $ee$ varies, though the box of 'reuse solution from previous step' is unchecked under the node of 'parametric sweep'. This is important because the steady configuration can be sensitive to initial conditions when the system have multiple energetic minimum. 2) How to perturbe the solution of previous step when the box of 'resue solution from previous step' is checked? In modeling of a system with strong nonlinearity, it's hard for the model to directly evolve from an arbitrary initial condition into the realistic final configuration in a single time-dependent study step, for the final configuration may depend on the intermediate process. So, I suppose the 'reuse solution from previous step' is needed. Neverthless, only reusing previous results can lead to trivial solution for all parameters. In the compressed bar example, with 'reuse solution from previous step', comsol keeps giving the trivial solution as $ee$ increases far beyond its critical value (shown below), maybe because the system only has trivial solution at small $ee$ and this trivial solution keeps being passed on to the next parametric sweep step as initial condition. As a result, the system is stucked in the trivial solution. Maybe a perturbation needs to be added into the previous solution before it is passed on to the next sweep step as initial condition. 3) How to suppress the storage of intermediate time solutions in each time-dependent study step during parametric sweep? In the psudo-dynamic approach, only the final solution at $t=T$ is realistic in a time-dependent study step. The intermediate solution from $t=0$ can be dropped, so that the disk space can be released. Is it possible to evaluate and store the strain energy for each $t$, so that I can assess the convergence, and meanwhile drop the rest solution before $t$ reaches $T$? I try to modify the output time list under the 'step1: time dependent' node, but sometimes this cause divergence.

By the way, I am using the Weak Form PDE module.

Thank you so much for your reading. I would appreciate it if anyone could shed lights on these questions.

Best regards,

HC L

How to Update Conditions Between Iterations

Mon, 29 Apr 2024 11:53:13 +0000

Hello,

I want to simulate the heat transfer over time, where I want to update the heat source condition in the model based on the extracted temperature data and where I want to apply the condition that if, during the simulation, the temperature rises from a certain point. The heat in the second region is automatically applied during that single simulation. If not, then that would be fine.

However, the main question or problem is how to apply this second heat transfer source, which depends on the temperature from the real-time simulation; during the simulation, this needs to store and instantly update the values, compare them with the condition by the next iteration, update the parameter depending on the condition in the next iteration, and get the results from it.

So, if someone guide me with this or even have small tip on how to do this then would be reallly greatful for me to solve my problem within the model.

Thank you so much in advance for your help.

gas sensing on semiconductor surface

Mon, 29 Apr 2024 11:43:15 +0000

Dear sir, How charge get transfer between analyte gas and semiconductor surface using COMSOL for gas sensing and how to couple this with semiconductor physics.

Error: "Failed to find a solution. Maximum number of Newton iterations reached..."

Sun, 28 Apr 2024 07:48:28 +0000

I'm trying to model a magnetic coil with magnetic shielding around it. I'm incorporating magnetic saturation into the model and I'm trying to model the working in this limit. My study is in the stationary domain, but I keep running into this error and I'm not sure if there is something wrong with the physical model I have or the way I'm going about it. I have played around with the Mesh size and the max Newtonian iterations, but nothing seems to work. Any help would be greatly appreciated.

I have the BH curve data interpolated as well. The file is too big, so I'm not able to attach it to this post.

mesh of rigid plane in contact mechanics

Sun, 28 Apr 2024 03:33:35 +0000

As attached a rectangular plane is drawn. Though the rectangular plane is not specified by the user to be a rigid body, it is assumed by comsol to be a rigid body. Does the mesh of the rectangular plane affect the simulation at all?

Microwave Heating with Rotating Mesh in COMSOL Multiphysics 5.3

Sun, 28 Apr 2024 16:45:21 +0000

Hello! I am trying to simulate the heating of a rotating object due to microwave heating. Therefore, I tried to use the heat transfer in solids model coupled with the eletromagnetic waves model and the rotating machinary with fluid flow. By this, the rotation apparently occurs, however, the results were the same as if the object were static. So, I suspect that the rotating machinary model isn't coupled with the microwave heating. Can someone help me with this problem, please?

About the 'Reuse Solution from Previous Step' in Parametric Sweep and Time-dependent solver

Sun, 28 Apr 2024 03:03:37 +0000

Hi there,

During the trail run, I encounter several questions:

1) The system doesn't seem to start from the initial conditions. For example, the result of $w$ at $ee=0, t=0s$ (shown belown) is quite different from the prescribed initial condition. And the $w$ profile at $t=0$ changes as $ee$ varies, though the box of 'reuse solution from previous step' is unchecked under the node of 'parametric sweep'. This is important because the steady configuration can be sensitive to initial conditions when the system have multiple energetic minimum. 2) How to perturbe the solution of previous step when the box of 'resue solution from previous step' is checked? In modeling a system with strong nonlinearity, it's hard for the model to directly evolve from an arbitrary initial condition into the realistic final configuration in a single time-dependent study step, for the final configuration may depend on the intermediate process. So, I suppose the 'reuse solution from previous step' is needed. Neverthless, only reusing previous results can lead to another question. In the compressed bar example, with 'reuse solution from previous step', comsol keeps giving the trivial solution as $ee$ increases far beyond its critical value (shown below), maybe because the system only has trivial solution at small $ee$ and this trivial solution keeps being passed on to the next parametric sweep step as initial condition. As a result, the system is stucked in the trivial solution. Maybe a perturbation needs to be added into the previous solution before it is passed on to the next sweep step as initial condition. 3) How to suppress the storage of intermediate time solutions in each time-dependent study step during parametric sweep? In the psudo-dynamic approach, only the final solution at $t=T$ is realistic in a time-dependent study step. The intermediate solution from $t=0$ can be dropped, so that the disk space can be released. Is it possible to evaluate and store the strain energy for each $t$, so that I can assess the convergence, and meanwhile drop the rest solution before $t$ reaches $T$? I try to modify the output time list under the 'step1: time dependent' node, but sometimes this cause divergence.

By the way, I am using the Weak Form PDE module.

Thank you so much for your reading. I would appreciate it if anyone could shed lights on these questions.

Best regards,

HC L

Coupling semiconductor and electrostatics physics in different domains

Sat, 27 Apr 2024 14:12:38 +0000

Hi. I am trying to simulate MOSFET with ferroelectric gate using Semiconductor Module. I am using semiconductor physics to simulate semiconductor part, but I need to use my own electrostatics equations (created with General form PDE) to interface with LGD (Landau-Ginzburg-Devonshire) equation in ferroelectric (dielectric) insulator part. How do I interface or couple semi with electrostatics? I need to couple electric potential from ferroelectric domain (with electrostatics equation) with semiconductor domain (with semi equation).

I tried many different setups. I think its something trivial that I am just missing. Any boundary condition? Making the semi and electrostatics with the same dependant variable (field name)?

I am sending my mph file, if anyone wants to looks at my model. I am thankfull for everyone who tries to help.

Species adsorption in Electrocoagulation

Sat, 27 Apr 2024 13:20:44 +0000

I am trying to simulate the electrocoagulation. And I have added the laminar flow, Secondary current distribution and also the transport of diluted species. The aluminum is getting dissolved into the solvent. and the pollutant is also getting dissolved. concentration of both species are increasing with time. I want to know 2 things. 1) How to go forward with the electrolysis at the cathode 2) How to simulate the adsorption of pollutant by another species

I want to know how to model Langmuir isotherm if possible.