AxisVM | Erdbebenberechnung, Teil 1 [de] — Transcript
Full transcript
- 0:00I welcome you to the webinar on seismic
- 0:06analysis using the response spectrum
- 0:08method. My name is Daniel Gas. I am
- 0:14supported today by Ms. Gabriel Müller
- 0:16and Mr. Dirk Menenger, who will be
- 0:17happy to answer your questions via the
- 0:19webinar panel on the right side of your
- 0:22screen. Today, we will focus
- 0:26exclusively on the actual seismic
- 0:29calculation. We will hold another
- 0:33webinar on March 30th regarding the
- 0:35evaluation of the results. We typically
- 0:41apply the response spectrum method to a
- 0:43spatial building model for which we
- 0:45perform a modal analysis. The standard
- 0:52stipulates that we should consider
- 0:54enough eigenmodes so that the sum of
- 0:56the modal masses—that is, the masses
- 0:59of the single-degree-of-freedom systems
- 1:01shown here—amounts to at least 90%of
- 1:03the total building mass. I will say
- 1:10more about that later. Based on the
- 1:16site parameters, namely the seismic
- 1:18zone and the soil class, we select a
- 1:21response spectrum. From this response
- 1:29spectrum and other structure-dependent
- 1:31parameters, such as the importance
- 1:33factor and behavior factor, we
- 1:34determine the design spectrum. Using
- 1:43the previously calculated natural
- 1:45frequencies, we enter this design
- 1:47spectrum to determine the maximum
- 1:48accelerations of the modal
- 1:50single-degree-of-freedom systems. By
- 1:57reversing the modal analysis, we obtain
- 1:59accelerations for the entire model. We
- 2:05can multiply these accelerations by the
- 2:07masses to obtain the seismic forces.
- 2:14Finally, we perform a static analysis
- 2:16taking these seismic forces into
- 2:18account and design the structural
- 2:20components. To implement this in Axis
- 2:29VM, we start in the Loads tab and first
- 2:32create a load combination. We will use
- 2:39this load combination later to
- 2:41determine the masses. We can name this
- 2:49load combination whatever we like. The
- 2:54type we select does not fundamentally
- 2:56matter. It only serves to group the
- 3:01load combinations if we create several.
- 3:08Technically, a user-defined combination
- 3:10is correct. We consider all permanent
- 3:16loads with a factor of 1.0 and variable
- 3:21loads with the factor PSI 2. This means
- 3:31we consider all permanent loads at
- 3:33their characteristic value and variable
- 3:35loads at their quasi-permanent portion.
- 3:44We neglect all loads here that are not
- 3:47caused by masses, e.g., wind loads, or
- 3:49that do not have a quasi-permanent
- 3:51portion. In many regions, this refers
- 3:59to snow loads, for example. We can now
- 4:07start the vibration analysis and, in
- 4:15the calculation parameters, select the
- 4:18load combination we just created and
- 4:20convert these loads into masses.
- 4:28Unfortunately, we cannot determine the
- 4:30number of vibration modes required in
- 4:32advance. We must make an assumption.
- 4:38Nine eigenmodes are set as the default
- 4:41value. I will leave it as it is. On the
- 4:49right side, we have further options.
- 4:55For one, we can neglect certain mass
- 4:57components. In this case, as in very
- 5:02many cases, I would like to neglect the
- 5:04vertical vibration component. I am
- 5:09therefore deactivating the mass
- 5:11component MZ. At the very bottom, I
- 5:19select which masses should be
- 5:21considered. Uh, I can consider all
- 5:26masses. That is the default setting. I
- 5:31can consider masses starting from a
- 5:33certain ratio. Or if I work with the
- 5:40floor identification at the top left,
- 5:42uh, I can select the masses above which
- 5:44floor, uh, I would like to consider. By
- 5:55considering or neglecting masses below
- 5:57a certain ratio, I define the clamping
- 5:59horizon. This means my entire building
- 6:06mass is calculated above the clamping
- 6:09horizon. With this, I can start the
- 6:16vibration calculation. In this case, it
- 6:20takes about one and a half minutes,
- 6:22which I would like to skip. I will
- 6:28therefore open another model. We have
- 6:35the results of the vibration analysis
- 6:37here. As result values, we receive
- 6:44exclusively deformations. These
- 6:48deformations are dimensionless. You see
- 6:52no unit in the coordinate window.
- 6:57Natural vibrations do not provide us
- 6:59with an absolute displacement. They
- 7:03only provide us with ratios between the
- 7:06displacement of individual nodes. We
- 7:10have not defined an absolute magnitude
- 7:13of excitation. Therefore, no absolute
- 7:17displacement can be determined. After
- 7:25the actual earthquake calculation, we
- 7:27will obtain deformations that do indeed
- 7:29have a dimension. In the very first
- 7:41step, I want to display the deformed
- 7:43model to verify these natural modes. I
- 7:52increase the magnification factor to
- 7:54see better how the building deforms.
- 8:03The first natural mode, as Axis VM
- 8:05designates it, is the first natural
- 8:08mode according to physical definition.
- 8:13That is, it is the natural mode with
- 8:16the lowest natural frequency, in this
- 8:18case 3.4 Hz. I can look at the other
- 8:24vibration modes. And I can consider, at
- 8:33least qualitatively, whether these
- 8:35natural vibrations are plausible,
- 8:37whether the building can even vibrate
- 8:39this way. I am now opening the table
- 8:49view. The table for the mass
- 8:54participation of the natural modes is
- 8:56opened automatically. We see here the
- 9:01share of the modal mass in the total
- 9:03building mass for each natural mode. As
- 9:11mentioned, we should reach at least 90%
- 9:14, which means 0.9 in this case. I have
- 9:21the option to switch off natural modes.
- 9:27I do not need them to reach these 90%.
- 9:33Thus, no earthquake loads are
- 9:35calculated for the inactive natural
- 9:37modes. I can thereby reduce the
- 9:41computational effort somewhat. In this
- 9:47respect, the standard differs from the
- 9:49Eurocode. The standard stipulates that
- 9:56we should reach this 90%modal mass
- 9:58participation without further
- 10:00conditions. The Eurocode goes a step
- 10:11further here and says we should reach
- 10:13this 90%modal mass participation, or at
- 10:16least consider all natural modes that
- 10:18individually have a modal mass
- 10:20participation of more than 5%. Uh, that
- 10:31is hard to verify if we should not
- 10:34reach this 90%modal mass participation.
- 10:41Uh, we do not know what shares the 12th
- 10:44or 15th natural mode might possess. Uh,
- 10:52but we can assume that these modal mass
- 10:54shares tend to decrease toward higher
- 10:56natural modes. It is not a steady
- 11:02decline, but the trend is there. If we
- 11:11have calculated 20 eigenmodes and we
- 11:13find that from the tenth eigenmode
- 11:15onwards no relevant modal mass
- 11:17participation is reached, we can assume
- 11:19, uh, that we will not have any
- 11:21eigenmode with more than a 5%share. The
- 11:31Eurocode establishes a second condition
- 11:33. If, uh, if we cannot meet this first
- 11:40condition, we should at least consider
- 11:43all eigenmodes that have an
- 11:45eigenfrequency below 5 Hz, uh. As a
- 11:53minimum, however, we should consider
- 11:55the square root of the number of floors
- 11:58in eigenmodes. You see the eighth and
- 12:05ninth eigenmode here in the eight. Here
- 12:08the eigenfrequency is displayed in red.
- 12:16This is more or less typical for higher
- 12:18eigenmodes and means nothing other than
- 12:20that the numerical error in the
- 12:22calculation of this eigenmode is
- 12:24greater than 10 to the power of -5.
- 12:34These eigenmodes are no longer 100%
- 12:37reliable. We have a relatively large
- 12:43possible deviation here due to the
- 12:45iterative calculation. I return to the
- 12:54load definition and under load cases
- 13:01and load case groups I can create an
- 13:03earthquake load case. In this
- 13:11earthquake load case, I can define the
- 13:13earthquake load. I must save the model
- 13:19first. and now have the option to set
- 13:24the earthquake parameters. I select the
- 13:31subsoil class I want on the left side.
- 13:37I define the design value of the
- 13:39horizontal ground acceleration. I
- 13:44define the behavior factor. I can enter
- 13:50the behavior factor twice. Once the
- 13:56behavior factor for calculating
- 13:58internal forces and, separately, a
- 14:00behavior factor for calculating
- 14:02deformation. The standards, both SIA
- 14:08and Eurocode, distinguish between these
- 14:11two values. or not. You can use the
- 14:15same value for both. Next, I define the
- 14:22, uh, importance factor and I get, uh,
- 14:29the design spectrum. Important, uh, to
- 14:34mention: it is the design spectrum, not
- 14:36the response spectrum. Uh, compared to
- 14:44the SIA standard, there is a difference
- 14:46, uh, in this representation, or rather
- 14:48two differences. The standard
- 14:54normalizes the time axis
- 14:56logarithmically or uses a logarithmic
- 14:58time axis. This changes the, uh, image
- 15:04of the design spectrum. And you see the
- 15:11note here: the design spectrum, as
- 15:13shown here, is not normalized to
- 15:15gravity. This means that compared to
- 15:23the design spectra you calculate
- 15:25according to SIA 261, you get a
- 15:27difference of a factor of 10. When I
- 15:39confirm the entry of this, uh,
- 15:41earthquake load, I receive a warning.
- 15:47The sum of the modal mass participation
- 15:49in the Z-direction is less than 90%.
- 15:56This is because we, uh, have neglected
- 15:58the, uh, vertical vibration component.
- 16:05I can ignore this warning and carry out
- 16:07the earthquake calculation in exactly
- 16:09the same way. I will go back to the
- 16:15earthquake parameters once more. I have
- 16:20a few other options here that I would
- 16:21like to show. If I have calculated the
- 16:29vertical vibration component, I have
- 16:31the option to consider the vertical
- 16:33accelerations. I can independently
- 16:39define the earthquake parameters here.
- 16:47You can refer to the code for how to
- 16:49determine the earthquake parameters for
- 16:51vertical seismic action. Under the
- 17:02Torsion Effect tab, I have the option
- 17:05to apply torsion due to accidental
- 17:07eccentricity. This means the seismic
- 17:14loads are shifted by 5%of the building
- 17:16width, once to the left and once to the
- 17:18right. Uh. For this, I must specify the
- 17:27levels at which the individual floor
- 17:29slabs are located. If I am using story
- 17:37detection, the individual stories are
- 17:39already listed here. Which I should
- 17:44still do. I can delete the stories
- 17:51below the base level, so the lowest
- 17:53story represents the level of my base.
- 18:03If I am not using story detection, this
- 18:05window is empty by default. I have the
- 18:12option to enter the levels manually or
- 18:14import them from the model. In this
- 18:18case, I define one point at each height
- 18:24of a floor slab. In doing so, I ignore
- 18:30these stair landings, for example. They
- 18:35do not represent a full story in terms
- 18:37of seismic calculation. Whether I want
- 18:43to specify the slab for this elevator
- 18:45overrun or not is essentially up to me.
- 18:53In the last tab, I can further adjust
- 18:55the calculation. The combination of
- 19:02modal responses refers to the
- 19:03combination of individual natural
- 19:05vibrations. According to the code, both
- 19:15SIA and Eurocode specify the SRSS
- 19:17method, which stands for Square Root of
- 19:19Sum of Squares. We have a second option
- 19:30. We can combine the earthquake or
- 19:36natural modes using CQC, the Complete
- 19:39Quadratic Combination. In this method,
- 19:47each natural mode is combined with
- 19:49every other one using an additional
- 19:51factor. You can find the definition of
- 19:56this factor in the Axis VM
- 19:58documentation. This CQC method has a
- 20:05clear advantage over SRSS. It provides
- 20:12significantly more realistic results in
- 20:14the case of dependent natural modes.
- 20:19Dependent natural modes are natural
- 20:21modes with very similar natural
- 20:23frequencies. If the natural frequencies
- 20:29of two natural modes are close to each
- 20:31other, it is practically impossible to
- 20:34excite one mode on its own. The other
- 20:40natural mode will always vibrate
- 20:41slightly as well. The default setting
- 20:48is automatic. This means the program
- 20:52checks if dependent natural modes exist
- 20:54. If so, the CQC method is used;
- 20:58otherwise, it is combined according to
- 21:01SRSS. It is also worth noting that if
- 21:08the natural modes are completely
- 21:10independent—meaning the natural
- 21:12frequencies are theoretically
- 21:14infinitely far apart—both methods are
- 21:16absolutely identical. In the lower
- 21:24section, we combine the components of
- 21:26the seismic excitation. This means we
- 21:29combine the directions. According to
- 21:34SIA, the default is again the SRSS
- 21:37method. However, you can also choose
- 21:42the so-called 30%rule. We apply the
- 21:48earthquake action at 100%in one
- 21:50direction and 30%in each of the other
- 21:52two directions. This combination method
- 22:00originates from the DIN standard.
- 22:05According to the Eurocode, both
- 22:07combination methods are generally
- 22:09permitted. Now, the earthquake loads
- 22:20have been calculated. We have received
- 22:25a large number of load cases. These
- 22:30four load cases, ERa to ERb. System
- 22:36load cases. We do not see any loads in
- 22:39these load cases. They are used to
- 22:44combine the individual earthquake loads
- 22:46. We obtain the result of this
- 22:53combination in these load cases, ER1 +
- 22:56to ER4-. We also do not see any results
- 23:03in these load cases. These load cases
- 23:07do not contain any loads in the actual
- 23:09sense. However, they will contain
- 23:12results later on. If we neglect the
- 23:21torsion due to accidental eccentricity,
- 23:24we would not have eight load cases here
- 23:26, but only two. They are called ER +
- 23:31and ER-. The remaining load cases
- 23:38contain the loads from the individual
- 23:40natural vibrations. We can see loads
- 23:47here. We can also label them, and you
- 23:53will notice that these earthquake loads
- 23:55are very small. We are in the order of
- 24:02magnitude of half a kilonewton to one
- 24:04kilonewton per individual load. That is
- 24:09absolutely normal. We receive a
- 24:14multitude of loads across a multitude
- 24:16of load cases. In the end, it still
- 24:21results in a significant earthquake
- 24:23load. The load case I am looking at
- 24:31right now contains the earthquake loads
- 24:33from the second natural vibration due
- 24:35to the mass component in the
- 24:36X-direction. Due to an asymmetry in the
- 24:45floor plan—that is, due to torsion
- 24:47resulting from actual eccentricity—I
- 24:50can also obtain earthquake loads in the
- 24:52Y-direction from the mass component in
- 24:54the X-direction. Finally, I will now
- 25:09perform a static calculation and I will
- 25:15obtain the results from the earthquake
- 25:16analysis. I would like to skip that as
- 25:22well. The static calculation takes
- 25:27about 30 to 40 seconds in this example.
- 25:36Basically, we are already finished with
- 25:38the earthquake analysis now and can
- 25:40evaluate the results. What is also
- 25:49important to mention is that when we
- 25:51perform the static calculation, we must
- 25:54choose a linear static analysis.
- 26:01Unfortunately, it is not possible to
- 26:03perform a non-linear earthquake
- 26:05analysis using the response spectrum
- 26:07method. This means we also do not have
- 26:15the option to consider non-linearities,
- 26:17such as tension cut-off. I will go into
- 26:25more detail about that in the second
- 26:27part of this webinar. Well, the
- 26:42governing results we receive here
- 26:43include both the results of the static
- 26:46calculation and the results of the
- 26:47earthquake analysis. I would now like
- 27:01to go into a few details. First, I
- 27:07would like to say something about the
- 27:09model itself. In this case, I entered
- 27:16the model just as I would for a static
- 27:18calculation. It is important that I set
- 27:26the hinge conditions realistically.
- 27:32This means that for these masonry walls
- 27:34, I have defined a moment release. This
- 27:43means I have fixed all displacements
- 27:45and released all rotations, which
- 27:47allows the masonry walls to transfer
- 27:49horizontal forces as well. If I do not
- 27:58want that, I release the displacement
- 28:00along the hinge line, meaning along the
- 28:02top of the wall. This means I set the
- 28:09stiffness kx to 0. This way, the slab
- 28:16can slide along the wall without
- 28:18resistance, and I do not transfer any
- 28:20horizontal forces to the wall. It is
- 28:27important that I make this adjustment
- 28:28either at the top or at the bottom. If
- 28:35I release this connection at both the
- 28:36top and the bottom, my wall will shift
- 28:38through the building without resistance
- 28:40. And the model becomes unstable. This
- 28:47means that in the end, the calculation
- 28:49is no longer possible. It is also
- 28:58important to correctly model the
- 28:59connections, especially for these stair
- 29:01flights. Such a stair flight represents
- 29:09a diagonal connection between two
- 29:11floors. In this case, it is indirect,
- 29:15via the stair landing. And I get a sort
- 29:21of wind bracing that can stiffen the
- 29:22building quite significantly. In this
- 29:30case, I have also defined a moment
- 29:32hinge at the top and bottom. And on the
- 29:40side, I have defined an edge hinge
- 29:41without significant stiffnesses. This
- 29:54means I have mostly set the stiffnesses
- 29:57to 0, and for the stiffness KY, meaning
- 29:59the transmission of forces in the plane
- 30:02of the stair flight, I have set a
- 30:04minimum stiffness of 100 kN per meter
- 30:06per meter. This stiffness is so small
- 30:17that I can consider the connection as
- 30:19structurally non-existent. It is not
- 30:26possible to define edge hinges without
- 30:28any stiffness at all. Therefore, I must
- 30:33provide a minimum value somewhere.
- 30:38Essentially, however, with this edge
- 30:40hinge, I am modeling a complete
- 30:42separation between the wall and the
- 30:44stair flight. I would like to add
- 30:52something else regarding the
- 30:54stiffnesses. It is crucial that we
- 30:58model realistic stiffnesses for the
- 31:00seismic calculation. We previously
- 31:07performed the seismic calculation using
- 31:09the same stiffnesses we use for static
- 31:11analysis. In the settings menu, however
- 31:19, I also have the option to define a
- 31:22stiffness reduction. Here I can specify
- 31:31factors to reduce the stiffness of
- 31:33individual structural elements. In the
- 31:38vast majority of cases, the stiffness
- 31:40of the walls is the deciding factor.
- 31:44This means it has by far the greatest
- 31:47influence. Which value you use here is
- 31:53more or less up to you. The standards
- 31:59do not specify a fixed stiffness
- 32:01reduction. In the literature, you will
- 32:07generally find values here between
- 32:10approximately 0.5 and nearly 1.0. I am
- 32:18choosing a stiffness reduction of 0.7
- 32:20here, as proposed by Professor Bachmann
- 32:27from the ETH in Zurich. When I now
- 32:37perform the natural vibration
- 32:39calculation, it is absolutely essential
- 32:41that I also take this reduced stiffness
- 32:43into account. As soon as I set the
- 32:51stiffness reduction, it is active by
- 32:53default. However, I highly recommend
- 33:00that you check this again. If you
- 33:05select original stiffness, the
- 33:07stiffness reduction is neglected. In
- 33:25this example, we have set the restraint
- 33:27level at the ceiling height above the
- 33:29basement. The restraint level is
- 33:38usually set at the height of the ground
- 33:39level. This means we neglect all loads
- 33:52from the basement and also the loads or
- 33:54masses acting directly on the basement
- 33:56floors. I have tried to represent this
- 34:06with colors here. This building has a
- 34:12basement that is half above ground
- 34:14level on one side. We can represent a
- 34:21restraint level not only as a
- 34:23horizontal plane. I would also like to
- 34:28show you right away how to do that we
- 34:37open the Load Cases and Load Case
- 34:39Groups dialog and we separate the load
- 34:45cases. into a normal self-weight or
- 34:53permanent load case and permanent loads
- 34:55below the restraint level. And we do
- 35:02exactly the same for the variable loads
- 35:04. I am hiding everything, I am hiding
- 35:18all components that are located above
- 35:24the restraint level. In this case, it
- 35:31is very easy because I have separated
- 35:33the material definition. I can simply
- 35:39select the material definition of the
- 35:41basement. Another possibility would be
- 35:48to mark the corresponding components
- 35:50and use this function to display only
- 35:52selected elements. I can hide
- 35:59everything else. I open the Property
- 36:12Editor in the Window menu on the second
- 36:18screen. I show the loads and I can, for
- 36:28example, select these loads with
- 36:30Control A and I can move all loads to
- 36:39the second load case. I do exactly the
- 36:45same with the variable loads. I mark
- 36:52all loads and I move them to the load
- 36:54case. Q, variable loads below horizon.
- 37:09In my load combination, I set these
- 37:11load cases to 0. This means I manually
- 37:19neglect the masses below the restraint
- 37:21level. It is important that when
- 37:27calculating natural vibrations I choose
- 37:34the setting "all masses" for the
- 37:36considered masses. I have manually
- 37:40neglected the masses below the
- 37:42restraint level. In the load
- 37:46combination, I leave this setting
- 37:51unchanged. the masses of these cut-off
- 37:56wall sections are still not considered.
- 38:09If everything works, then essentially
- 38:11everything has been said about seismic
- 38:14analysis. However, I would like to show
- 38:19you something that can go wrong with
- 38:21vibration analysis. I have supplemented
- 38:27the model here with wall pieces in the
- 38:33stairwell. In this case, local natural
- 38:41vibrations occur. We have such an
- 38:45example here. In this case, only this
- 38:49stairwell wall is vibrating. The rest
- 38:56of the building essentially does not
- 38:58move. These mode shapes are not
- 39:05relevant for seismic analysis. You can
- 39:15usually recognize this very well in the
- 39:17deformed view, as these local natural
- 39:19vibrations typically produce large
- 39:21deformations in these vibrating
- 39:23components. Under certain circumstances
- 39:31, you can highlight this even better
- 39:33with the magnification factor. In the
- 39:42table of mass participation factors,
- 39:44you can easily identify local mode
- 39:46shapes because the modal mass
- 39:47participation factors are very small.
- 39:57You may also get other mode shapes with
- 40:00small modal mass participation factors,
- 40:02but local mode shapes typically have
- 40:04small mass participation. Unfortunately
- 40:11, also typical—and this is the
- 40:13unpleasant characteristic—local mode
- 40:19shapes very often appear in clusters.
- 40:25This means that several local mode
- 40:27shapes have nearly the same natural
- 40:29frequency. If we look at this here, you
- 40:39can see it quite well; various
- 40:41vibrations of this stairwell wall occur
- 40:44at almost—or, down to two decimal
- 40:46places, even exactly—the same natural
- 40:49frequency. We could basically just
- 41:03ignore these local vibrations, but then
- 41:05we no longer reach the required 90%mass
- 41:08participation. In this case, I wouldn't
- 41:17worry too much about it. We are 5%below
- 41:23this threshold in the Y direction. Uh.
- 41:34Considering that seismic calculation
- 41:36does not have the same precision as a
- 41:39static calculation anyway, I can
- 41:41conclude that 85%is greater than 90%.
- 41:50But it can become more extreme, to the
- 41:52point where I would no longer want to
- 41:53allow it. The simplest approach to deal
- 41:59with this is to calculate more mode
- 42:01shapes. However, I might need a very
- 42:07large number of mode shapes to handle
- 42:10these local vibrations or to avoid the
- 42:16influence of these local vibrations.
- 42:22Another approach is to exclude these
- 42:24local vibrations. To do this, I go to
- 42:33the Loads tab. I select this wall.
- 42:45Let's do this the simple way. Ah, it is
- 42:49not a different material either.
- 42:53Therefore, I have to select the areas
- 42:56individually. I open the property
- 43:07editor again and shift this load as
- 43:09well. Below the restraint level, so to
- 43:20speak. This means I am neglecting the
- 43:26loads on the affected components.
- 43:44Neglecting the load or assigning the
- 43:46mass of a single wall doesn't change
- 43:48much in the natural vibrations. But if
- 43:56I neglect the masses of all walls, for
- 43:58example, then I am making a not
- 44:00insignificant error. I can represent
- 44:07this mass again by creating an
- 44:15additional load case. For this, I
- 44:20choose an independent load case.
- 44:26Independent load cases are not
- 44:27automatically combined with the other
- 44:29load cases after the static calculation
- 44:31. And now I have the option to apply
- 44:40the wall weight at the wall base and
- 44:46wall head as a line load and include it
- 44:58in the load combination for mass
- 45:00determination. This way, the total mass
- 45:12we represent is correct, but the
- 45:14individual walls that cause local
- 45:15vibrations are considered massless and
- 45:18therefore can no longer vibrate on
- 45:20their own. The local mode shapes no
- 45:28longer occur. Finally, I would like to
- 45:39make a single remark regarding
- 45:40calculation time. The calculation time
- 45:47is influenced by various factors. It
- 45:53stands to reason that computation time
- 45:55depends heavily on the model. The
- 46:01larger the model, the longer the
- 46:03calculation will take. Uh. However, the
- 46:12calculation depends not only on the
- 46:14size but also on the complexity of the
- 46:16model. I recommend simplifying the
- 46:21model as much as possible. This means,
- 46:26for instance, if two walls are almost
- 46:29touching, I would model them as one.
- 46:37This could be the case if the wall
- 46:39thickness changes and the outer edge is
- 46:41flush; I wouldn't model the wall axes
- 46:43offset. Mesh generation also logically
- 46:52influences the calculation time. I
- 46:58tried this out using this example. I
- 47:05obtained very different calculation
- 47:06times for different mesh sizes. For
- 47:14instance, with a mesh size of half a
- 47:16meter, I had to wait almost 13 minutes.
- 47:24I wouldn't necessarily recommend this
- 47:25mesh size to you. It is relatively fine
- 47:31, at least for an entire building. It
- 47:37can make sense to mesh individual small
- 47:39components with half a meter. On the
- 47:49other hand, with a 2-meter mesh, I get
- 47:51just over a minute of calculation time,
- 47:54and with the maximum possible mesh size
- 47:56, I'm under half a minute for the
- 47:58vibration analysis. Uh, maximum
- 48:05possible. That means, I think I entered
- 48:08about 50 meters. Due to the constraints
- 48:16at the corners of the components, such
- 48:18a large mesh is not even possible. With
- 48:27an optimized mesh, that is, with
- 48:29different mesh sizes depending on the
- 48:31dimensions of the components. In this
- 48:37case, between half a meter and I
- 48:39believe 1.80 meters, I get a
- 48:41calculation time of 1 minute and 40
- 48:43seconds. I consider that absolutely
- 48:50reasonable for such a building.
- 48:57Calculation time is logically also very
- 48:59dependent on the system environment.
- 49:06Excuse me. Essentially, calculation
- 49:12time is influenced by processor
- 49:14performance and, up to a certain size,
- 49:16by the RAM. I say up to a certain size;
- 49:24above 8 GB of RAM, there is hardly any
- 49:26noticeable impact. However, below 4 GB
- 49:34of RAM, the influence of the memory is
- 49:36relatively high. Another thing that
- 49:43plays a big role is where the model is
- 49:45saved. Calculations on network drives
- 49:51can be significantly slower than
- 49:52calculations on a local hard drive,
- 49:55depending on the network. I would
- 50:00recommend that you save the model
- 50:02locally, especially for earthquake
- 50:04analyses. As already mentioned, we will
- 50:11hold the second part of this webinar on
- 50:13Friday, March 30th. March, on the
- 50:19evaluation of earthquake loads; on
- 50:21February 23rd. February. we will insert
- 50:31another webinar on moving loads in
- 50:32between. I hope that we will see each
- 50:37other again on that occasion. I thank
- 50:42you for your attention and wish you a
- 50:44nice Friday.
About this transcript
This page contains the full transcript of AxisVM | Erdbebenberechnung, Teil 1 [de] by Ingware AG, generated from the public captions YouTube serves with the video. The transcript has 4,235 words across 670 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.