설명
주요 학습
- Learn about choosing the correct analysis type.
- Learn about assembly simplification.
- Learn about selecting the correct mesh type and adjusting the suitable settings based on the models.
- Learn about making adjustments to converge results to reality and editing the model after the analysis if it is required.
발표자
HAVVA SIMIT: Hello, everyone. Welcome to Autodesk University 2023, The Design and Make Conference. I'm Havva Simit. I'm a mechanical engineer. And today I'll be presenting Convergence of Results to Reality in Analysis Using Inventor Nastran.
Before presentation, I will talk about myself. I've been working at Prota, which is a platinum partner of Autodesk in Turkey. It's also an Arkance system company.
I've been working here since 2018, no '17, as a Consultancy Training and Technical Support Specialist. I have different levels of experience of Autodesk products. Mainly, I'm responsible for Inventor, Fusion 360, Vault, Nastran, and AutoCAD.
Today's topic is Convergence of Results to Reality in Analysis Using Inventor Nastran. Before starting the complex things, I just want to talk about why I choose this topic. Since at the end of the 2018, Nastran has been available in PDMC. We can find many tutorials about various analysis subjects on different online platforms. Almost all those tutorials shows you how to apply a lot, make contact, and et cetera.
But any of them tell us how to understand what is the result, how we can understand it, how we can improve the productivity of the system. What does it mean the maximum stress or the range of the stress? What we should understand about this information. Today, my goals are to make you understand why we do these steps, make you able to read the results, and see what you can change to improve the correctness of the results.
To specialize in analysis process, sometimes we have to know the production methods, make the right definitions like float, match types, and et cetera, and interpret the results well. Otherwise, the results will just consist of the numbers that do not mean anything. We all know how to input temperature, strength, information in Nastran. Before data entry, we must verify the physical data of the parts and assemblies included in the analysis.
At some points, we should know the production methods. And accordingly, we should have an idea about the materials behavior beforehand. These two things will enable us to understand how close our analysis, analysis results are to reality.
The examples of the topics I will talk about will mostly be based on linear and nonlinear static analysis, due to my previous experiences. To ensure the accuracy of the results, we can divide the control process into two main headings, as before and after analysis. Things to look at before solving the analysis, as you can see on the left side-- simplification of the model, control the physical properties, material behavior, correct input of data, such as force pressures, mesh definitions, and the last one, converging the settings. After solving the analysis, we should consider these things-- safety factor, distribution of the stress, displacement, and the effect of the mesh size on understanding the results.
Let's study on the process, starting with what needs to be done before solving the analysis. The first step is simplification of the model. In the simplification of the models, firstly, the parts that are not included in the system should be removed or simplified. And from now on, I'll continue with an example of a boiler to make all the items more understandable.
As you can see, the model consists of three holders to hold the boiler in the middle of the working space, the boiler itself, and the base part at the bottom to make the stem stable. If the connected pipes or parts at the top or bottom and external elements do not affect the analysis or analysis result, you should delete them or make all unrelated parts one piece. But what is the reason of the simplification?
Duration of the analysis depends on the number of elements, as you can see on the right side. But what is the element? On the right side, you can see the element numbers. The number of elements represents the mesh number.
As the model grows physically, the number of mesh increases. According to the physical growth, the analysis will be solved in a longer period than it should, because the solution time of the analysis depends on the size of the model and therefore, the number of meshes. Prolonged resolution means loss in both money and business. That's why we need simplification.
Second item is controlling the physical properties of an identified material. There are laws and materials defined by different names in the Inventor Material Library. The values of these materials that are important for us are yield strength, tensile strength, density, shear modal, sometimes young modal and Poisson's ratio. And if there is a temperature, the thermal conductivity coefficient value will be most important things in the physical properties.
You can do this by editing in the library section. Also, you can edit it from Nastran material assignment panel. When you edit Nastran Material Library, the Inventory Material Library will not be affected, but if you want to edit the Inventory Material Library permanently as a template, the Nastran Material Library will be affected. On the right side, a newly defined material will not be included in Inventory Material Library. It's just valid for Nastran.
And there's a thing we should pay attention to in the physical properties, temperature. As temperature increases, yield strength and elasticity modular modules are affected. As you can see, the yield stresses and elasticity modules decrease as the material is exposed to temperature.
Another important factor is the thickness of the metals. Depending on the thickness, yield and tensile stresses may vary. For this reason, before defining the material, the analysis, it's necessary to study its physical properties and make material assignments accordingly.
The third one is material behavior. When we come to the topic of the material behavior, we need to be mastery a very basic engineering graphic. This graphic is the stress TO strain graphic. So why do we need to know this?
There are two different types of analysis linear and nonlinear analysis. In linear analysis, as you can see above, the curve in the elasticity region increases linearly and breaks at the maximum stress. However, this curve is more suitable for materials like ceramic, plastic, glass, and et cetera. The toughness of these materials is low. That is, they break without stretching and without much energy.
Suppose a force to be applied to a part, and let this force be the pulling force. In the linear static analysis, we assume that the stress increases continuously, depending on the force. This, at the maximum stress-- where is my cursor? In this static analysis, we assume that the stress increases continuously from right here, depending the force.
Thus, at the maximum stress, we think that the part breaks without any plastic deformation. However in reality, it's not that true. Metal material is stretched a certain amount. And its shape is changed. It's plastic deformation, you can see the molecule's moment right here and bricks at the stress, below the maximum stress.
The shape-changing step is called plastic deformation. The stress-strain plot above show how a metal will deform when subjected to tension. As you can see, the graphic is nonlinear because metals are materials with a certain elongation coefficient value.
After passing the yield stress from right here, stress valley, the metal, metal is brick. By looking at the stress, the value is obtained after download is completed. We test how safe the deformation of the part is. Because the yield and tensile stresses of metal materials are certain, if there is no dynamic loading and it's not exposed to irregular factors such as heat or wind, we can decide whether the part is safe or not by comparing the analysis result with yield and tensile stresses.
If it's subjected to stress at the critical limit or above, the breaking stress, we can test the safety of the model by increasing the thickness and adding additional supports. Or we can change the whole shape of the model. For nonlinear analysis, there is a panel like you see on the left side. To define the type of material, these are elastic, elasto-plastic, and plastic on plastic.
The fourth one is correct input of data, such as force and pressures and all other loadings. When entering data such as force and load, the type of load point distributed load, preload, and et cetera, and its angle are important, especially for the correct value of the axial load. If it's not applied correctly, the analysis result can be understood completely different.
In the example, you see a 23 kilonewton preload is defined from right here for the bolted connection. On the material part to tighten the bolt, normally we calculated the maximum stress value is around 280 megapascal. However, when the analysis was completed, the maximum stress was determined at 2,317. Due to the preload value, it's an axial stress, like you see.
For this value's reference, this is, if the axial stress value is referenced, the system will not be considered safe. But when the system is examined in general and the measurement is made with virtual prop, it seemed that the values are well below 2,317 megapascal, which is the yield strength. It will be easier to understand the reliability of the system when we realize that this maximum stress is axial for this example.
The other step is mesh definitions. During analysis, we break the models into small pieces, which we call meshes. There are different types of it-- linear, triangular, and quadrant. These mesh structures divide the surfaces into small areas of the size we specify, allowing us to understand how much of an area the stress form on those surfaces affected.
While applying the mesh, three different mesh related data are entered. The first is in the idealization section. There are three types of it-- solid, shell, and line elements. You can measure model, including the inside of it. It's 3D, and we call it solid elements. This mesh type is mostly used for non-sheet metal parts.
The shell mesh is generally suitable for suitable for models such as the construction steel profiles and sheet metal parts. Interpreting the results for the sheet metal parts will be difficult because of the thin wall. In this situation, we prefer to use shell mesh type.
If stress accumulation is expected in a linear region, we can also add the linear mesh type to necessary corners with line elements. In some cases, general mesh size may not be enough for you to interpret the result correctly. Enlarging the overall mesh size will increase the number of elements, which means this extends the solution time. By externally assigning smaller-sized local mesh such as vertex, face, edge, or part types, you can make interpreting the results easy.
Finally after determining these definitions, you will create a mesh determining a mesh size. Nastran offers a scale of how well the size fits your fits your mesh. But at this point, you need to know something important. Choosing the right mesh size comes with experience.
First, a mesh size is determined and the analysis result. Then the mesh size is increased or decreased a few times. The analysis results are compared. Then we can decide what is correct according to comparison.
Now, you see a model, which has been different. Mesh size and types include local mesh from right over here, right here, and at the bottom part. The bottom part mesh size is huge. It's mesh size is not important for us.
We are not interested in the stress of bottom part. We just need it to keep the system is stable. When we look at the boiler, its mesh size is smaller than the bottom part. We will examine the stress on the cylinder body.
This is a solid model, not phase or sheet metals. According to the results, this mesh size is proper for it. The last parts are holders. As you see, it has the smallest mesh size because of this its thickness. In a system, we understand that we understand the results well. We must define different sizes and types.
The last one is convergence settings. Finally, another settings we can use in analysis is converging settings. Here, you can see two types of refinements, global and local. Normally, not all meshes can be created 100% correctly when we create it creating the mesh structure.
At surface, transitions, corners, and et cetera, the size and shape of the mesh can change abruptly. In these cases, the system should be able to keep under the control of the mesh to be applied. With the local and global mesh refinement settings, you can limit how much the ratio of the consecutive mesh to each other can change. Use the mesh convergence settings to refine mesh settings during analysis, either globally for an entire part, or locally in areas of complexity with more mesh areas.
Here, you can enter the maximum number of refinements and what rate the mesh structure will be improved. The maximum number of refinements specifies the maximum number of each refinement cycles for convergence. Refinements can case before reaching this number. If the stop criteria is reached.
And what is the stop criteria? Stop criteria says the minimum percentage value for the difference between subsequent refinement results. If the value is less than this minimum percentage, the refinement stops. Results to converge are for one wellness of stress. Refinement factor controls the grow rate of the elements in critical regions for each refinement run.
Now, we are going to study on results. One of the most important outcome evaluation criteria is safety factor. Let's continue with an example to make it easy. Here are the safety factor results of the holder. We can expect to see the safety factor value bigger than 1.
So we can accept the holder stays safe. But most of the time, to stay in a safe area, 1 is not enough value. 1 and 1/2 and more is more acceptable. As you can see in the picture, the safety factor is always more than 1 and 1/2, except the exit point of the hole. The minimum safety factor is 0 here. But we said that before end preload is applied right over here. So its minimum value is 0, but it's not important because it's an axial load right over here.
The maximum safety factor is around 290. It means that this area does not require-- this blue area, it does not require any change, such as shape, material, and et cetera. However, the safety factor's axial point is not a problem for this study. We already know that the value is high because of the preload of the bolt.
What is important? We should consider the sharp corners like here, this area. And at the bottom, we should consider this corner. Those may need to be updated, like changing the shape or thickness of the model. Or we should use more suitable materials for the holders.
Let's move with a model first. The other one is distribution of the stress. And it should look with the model.
First, the maximum and minimum values are shown by Nastran, as you see right here. Other values are in a diagram, measuring the critical area, critical red areas, which probe will show us, what is the value of the stress at the measurement area.
Then you decide what is required for the shape changing. You see the stress concentration is higher in the corners and the axial points. You can make it fill it right here or redesign. Or you can make a smooth transition right here. Under developments, this part, there's a wetland right over here.
There are a few more overstressed areas. At those points, you can change the shape or thickness. Or you can use a more strengthful material. Or you can use something a shape like rib.
The third one is displacement. Every material has an elevation coefficient value. It's affected by loads, temperature, and external.
We need to evaluate this criteria, while examining these displacement areas from right over here. And displacement has no sort of value to evaluate. We must first calculate the amount of displacement by hand, according to the allocation coefficient value. When the result is bigger than the calculated value, as you see, it's 4.3mm, we may need to change the shape again or use more resistant material.
The last one is the effect of the mesh size on understanding the results. To see the distribution of the stress, we divide the models into small structures. The smaller dimension, the smaller the dimensions of these mesh structures, we can see how much the stress value affected the area. Let's say the mesh size was 5 millimeter. After the analysis, the stress seems like it affects a 20 square millimeter area. We decrease the mesh size into 2 millimeter. Then we solve the same analysis again.
After the solution, we saw that the stress affects just a 6 square millimeter area. After we decrease the mesh size a few times, if the results do not change notably, it means we found a suitable mesh size for the model. Let's move on to a practical example.
As you can see, we have an assembly consisting of three holders-- ground, boiler, and [INAUDIBLE]. Since the other elements of the system didn't have any effect on the analysis, I deleted them completely. When we move to the analysis environment, the first thing we need to do is defining the new material. Or if it's already in the Inventor Material Library, we can choose from right over here or here.
We need to define the materials first from right over here or edit and control the necessary data of the defined materials. As you can see, the use stress, tensile stress, Poisson ratio, shear modulus, elasticity modulus, thermal expansion coefficient value, they are important for us. We can edit the information as we wish from here, or assign it from the library if it is available in the inventory library, of course.
We can see the definitions of these materials in the section under the idealization. As you can see when we enter the idealization menu, you can assign the defined material to the relevant parts. From the assembly environment, right over here, we can choose the materials and then assign to the geometry right over here. After that, we can choose the idealization types, solid, shell, or line elements.
After selecting the general mesh structure from right over here, the idealization menu, we threw away the-- as you can see here, there's just-- just show me the connectors. As you can see, we define a preload, which is a 23,000 Newton right over here for the bolted connection. And we can define this loading as a movement by axial or a moment.
Then I fix under the base part. So the model, let me show you this. I'll fix the bottom part with the constraint. Fix comment-- there is no transition or rotation. Then I will define the loads.
The first one is Newton. Sorry, it's minus 718 Newton. The other load is pressure. It's defined in the middle of the boiler. You can see the red arrow right over here. Its value is 1.3 megapascal.
And the last one is body temperature. It's around 373 Kelvin. After defining the loads, we can move to the local meshes. As you can see from right over here-- as you can see, five different local mesh assignments have been made. You can assign these local matches to surface, vertex, or parts.
You can control stress concentration by dividing the parts related to local meshes into smaller pieces. And you can find out how much area they affected. When I added these local meshes, you see, I'll choose the part. And I will give the mesh size of it in millimeters unit. For all parts, five different local measures are defined at the mesh control size.
We define the local mesh. Then we will go through the mesh settings. I'm not able to load this window. I don't know why. Local mesh and general mass settings are important for converging settings before because there are optimization details for local and general mesh settings in the converging settings right over here. Therefore, local meshes are important when importing mesh structures and converting the results.
What we need to look at next is to examine the results after the analysis. And I'm not going to solve the analysis right now, because it takes around 4 or 5 hours. First let me change the window. That's OK.
First, I check the stress from right over here under the subcase results. I'll load the results. Then I will displace the Von Mises stress. This is the format.
As you can see, the stress concentration are greater at the sharp corners. I check the values right here with the probe to better understand them, with probe comment. Then look at the value and think about what I need to do into the model.
Here, I can design a smoother transition around the sharp corners right here in these areas. It seems like it's blue. But it's just because of the maximum axial force. I will change the range of Von Mises stress with the tensile strength, tensile stress, sorry. It's 370 megapascal. Then I will display the stress graphic.
Right here, it's more clear to understand. As you can see, the red areas, it's mostly around the sharp corners. But here, the maximum stress is 2,370 megapascals. However, this values exist axially in the pre-stress defined region. Since the values of other fields are closer to the strength, I limit the stress range value to between 0 and 370 megapascal and examine the model.
As you can see, the stress concentration are bigger right over here in this area and this any of the holders. I check the values here with the probe a few seconds ago. And I look at the value and think about what I need to make for these holders to improve its a strength.
I can redesign a smarter smoother transition right over here and here. Or I can increase the thickness of the holders. Or I could try working with a more resistant metal material. There is some stress accumulation in the area where this stress is defined. These are caused by crushing, due to the squeezing process.
Another value I check is the displacement value right over here. By the way, we can close the mesh structure from visibility options and lows or constraints. It will be more clear to understand.
Yes, here we must first calculate the allocation amount manually. Here, maximum value is around 5 millimeter. These are towards the top of the boiler. It's generally between-- the displacement value is around between 0 and 1mm in the holders in the top and the bottom of the holder in the boiler.
The reason why it is high in the upper region is due to both the temperatures defined inside and the general environmental temperature of 3,373 Kelvin. We can find out whether this 5 millimeter value is critical or not by comparing it with manual calculation. Another value is safety factor.
Well, these are checked by measure measuring critical areas with the probe. Then we will see, it's bigger than 1, you see. I will change the range between 0 and 2 or 10. It's not matter, because 600 is really big.
I'll display it again. As you see, the value we will check is the security coefficient or safety factor. Values are checked by measuring critical area with probe.
As you see, the, red area value is bigger than 1. As I said before, 1 and 1/2 is more suitable for us. But 1 is the minimum limits for us.
Then the necessary model adjustments are made by taking into account, the stress value in these regions. Actually, this was the study. I hope it was helpful to apply the analysis and interpret the results. For questions, you can reach me where my email address or LinkedIn account. Good luck with your studies. Let knowledge be your light in the future.