I know that external factors or errors in the construction of a laboratory model influence the experimental results. I modeled a dam with a numerical model (with FLAC-3D). There is a good agreement between this analysis and the experimental results. What is the compelling reason for the small differences? Also, the results are different for two different numerical models. Does only the difference in the governing equations affect this difference or is there another reason?
You can use the uncertainty modeling factor Xm. It is defined as the ratio of numerical results/ test results. Then, checking "Mean" and "COV%" values.
Experimental results are affected by the instrumental and random errors. Numerical methods are affected by the physical model selected and the parameters belong it. For example, when you are using a turbulence model some parameters can be modified and normally you only have general recommendations for do that. On the other hand, the used mesh is another source of variations in results. The Combination of mesh, models, adaption and convergence process affect the simulation. That is the reason why is so important validate numerical results against good experimental results always be possible.
@Sara Irani
Difference is due poor knowledge of theoretical modeling.
If the same experiment is repeated, and result (of experiment ) remains the same ,then one has to propose the model to fit .
Actually, there are many reasons for this difference such as boundary conditions, model parameters, loading conditions....etc. Also, is the physicals element test mirror stress conditions in the field?
@Marwan Khashila
In spite of all the secondary things ,one can
present a mathematical model that fits with the experimental result provided ones mathematical knowledge is good . Anyway poor knowledge in mathematical model
makes ones to work day and night . Hope ...
............
Biswanath Rath
Yes, for sure. I mean if she used a well established model and calibrated its parameters accurately, still different factor must be considered.
Thank you for your clarification .
Difference is due to sampling size, bias and parameterisation of the model :)
Andrew Paul McKenzie Pegman
Marwan Khashila
Biswanath Rath
Sorry for delaying my reply.
In fact, I modeled a nonlinear dynamic analysis on a Rockfill dam with Finn's behavioral model with a 10-meter mesh and hysteresis damping with FLAC-3D.
Your comments are very correct. In fact, I wanted to see if my opinions were reasonable.
I also concluded that: First, if the example I have used for verification is a laboratory and experimental study, factors such as human error and the scale of construction and accurate specifications of materials and etc, can lead to errors in the results of analysis and testing. And if the example we have considered for verification is a numerical analysis with the help of software, factors such as material specifications, differences in the behavioral model, differences in the software with which we analyze, differences in the type of mesh and mesh dimensions, Differences in selective damping, differences in the parameters of a damping obtained by calibrating experimental graphs, can all be effective as a result of analysis and difference with another study.
I model with different meshes (square mesh, triangular mesh, etc.), with different mesh dimensions (8 meters, 10 meters, 12 meters, 20 meters, etc.) with different behavioral models (Mohr-Columb, Finn and. ..) I modeled the hysteresis damping (with different formulations and different numbers for a formula with a difference of 0.001) and with different reservior conditions (empty reservior and with N.W.L). And the results were different for all of them.
So I think your feedback and the results I have got are quite convincing.
Thank you.
with respect.
I. About numerical simulation and experimental data ― Numerical models can be used to generate data ― numerical simulation ― which can be just as experimental as if generated by physical models. What seems of utmost importance is to precisely characterize the source of data; such as industrial data, pilot plant data, or laboratory data. What classifies data as experimental is not its source, but the compliance with the so-called experimental (scientific) method.
II. Application example ― The recursive least squares algorithm (RLS) allows for (real-time) dynamical application of least squares regression to time series. Years ago, while investigating adaptive control and energetic optimization of aerobic fermenters, I have applied the RLS algorithm with forgetting factor (RLS-FF) to estimate the parameters from the KLa correlation, used to predict the O2 gas-liquid mass-transfer, while giving increased weight to most recent data. Estimates were improved by imposing sinusoidal disturbance to air flow and agitation speed (manipulated variables). The power dissipated by agitation was accessed by a torque meter (pilot plant). The proposed (adaptive) control algorithm compared favourably with PID. Simulations assessed the effect of numerically generated white Gaussian noise (2-sigma truncated) and of first order delay. This investigation was reported at (MSc Thesis):
Thesis Controlo do Oxigénio Dissolvido em Fermentadores para Minimi...
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress