What is Machine Learning?
Machine Learning is an application of artificial intelligence (AI) that allows systems to learn and improve from experience without being explicitly programmed automatically. Machine learning focuses on developing computer programs that can access data and use it to learn for themselves, the learning process begins with observations or data, such as examples, direct experience, or instruction, to look for patterns in data and make better decisions in the future based on the examples we provide. The primary aim is to allow the computers to learn automatically without human intervention or assistance and adjust actions accordingly.
What is the Design of the Experiment (DOE)?
Design of Experiments (DOE), also referred to as Designed Experiments or Experimental Design, is the systematic procedure carried out under controlled conditions to discover an unknown effect, test or establish a hypothesis, or illustrate a known product, it involves determining the relationship between input factors affecting a process and its output. It helps to manage process inputs to optimize the result.
Sir Ronald A. Fisher first introduced the method in the 1920s and 1930s. Design of Experiment is a powerful data collection and analysis tool used in various experimental situations. It allows manipulating multiple input factors and determining their effect on the desired output (response). By changing numerous inputs simultaneously, DOE helps to identify significant interactions that may be missed when experimenting with only one factor at a time. We can investigate all possible combinations (full factorial) or only a portion of the possible combinations (fractional factorial).
A well-planned and executed experiment may provide a great deal of information about the effect on a response variable due to one or more factors. Many experiments involve holding certain factors constant and altering the levels of another variable. However, this “one factor at a time” (OFAT) approach to process knowledge is inefficient compared to changing multiple factor levels simultaneously.
As was mentioned previously, there are multiple approaches to DOE. OFAT, Full and Fractional factorial, Taguchi and Response surface methodology (RSM), however, the RSM proves to be the best approach among the existing methods for DOE.
What is Response Surface Methodology (RSM)?
Response Surface methodology, or RSM for short, is a set of mathematical methods determining the relationship between one or more response variables and several independent (studied) variables. This method was introduced by Box and Wilson in 1951 and is still used today as an experimental design tool. RSM is a set of statistical techniques and applied mathematics for building experimental models. RSM aims to optimize the response (output variable) affected by several independent variables (input variables).
An experiment is a series of tests called executions. In each experiment, we make changes to the input variables to determine the causes of the response variable’s changes.
In response level designs, constructing response procedure models is an iterative process, as soon as we obtain an approximate model, we teste by the good-fit method to see if the answer is satisfactory. If we don’t confirm the answer, then the estimation process starts again, and we perform more experiments, in designing experiments, the goal is to identify and analyze the variables affecting the outputs with the least number of experiments.
This method achieves the best response surface by discovering each design variable’s optimal response level. In designing experiments, the goal is to identify and analyze the variables affecting the outputs with the least number of experiments.
What is Optimization in ANSYS Fluent?
Optimizing a product is one of the most important stages of product production. In the past, trial and error strategies were used to produce more stable products with better performance. But today, with the development of technology and to reduce production costs, trial and error methods can’t be appropriate, and there is a need for more reliable methods to design and optimize products before the production stage. Today there are different software with different CFD methods available to engineers to design and analyze the model before making the real model and applying the desired optimizations.
Using ANSYS, one can perform two types of Optimization. 1-Direct, 2-Indirect. Direct Optimization predicts the behavior of a system without any intermediary step. In contrast, indirect Optimization needs the data obtained by the RSM to provide the user with the correct mathematical function for predicting the system behavior.
The MR-CFD team conducted numerous simulation projects for optimization applications. For instance, the product ‘Optimization of a Compressor Cascade Using MOGA (BBD DOE & GA RSM)’, simulates a Compressor Cascade with different input and output parameters using the BBD method and then determines the sensitivity of the output parameters to the input parameters in two types of diagrams using the genetic aggregation method. Finally ANSYS Workbench has suggested 3 points as optimal points.
In another project, we optimize a combustion chamber performance using DOE and RSM. The target of this project is to optimize the geometrical parameters of the combustion chamber for targets such as maximizing the value of heat generation rate while minimizing the amount of formed pollution. Two types of optimization are examined in this project; Indirect optimization using the RSM method, using the CCD method and direct optimization.
Furthermore we discuss the Optimization models in details and widely in our blogs, written by the experts of MR-CFD team. For instance in the ‘Practical Exercise Applying DOE & RSM for an Optimization’ blog, different types of DOE and RSM methods are explained and is able to give you an understanding about the concepts of Optimization by an exercise.
MR-CFD Machine Learning/Optimization Services
With several years of experience simulating a wide range of problems in various CFD fields using ANSYS Fluent software, the MR-CFD team is ready to offer extensive modeling, meshing, and simulation services. Simulation Services for Machine learning/Optimization are categorized as follows:
- Direct and Indirect optimization process for different fluidic systems
- ML models for different projects to predict the behavior of the related system
- Calculating different mathematical correlations for predicting the behavior of various types of systems
- Obtaining the optimal state of a process inside a system to reduce the overall costs
- Designing various equipment for maximum efficiency
You may find the related products to the Machine Learning/Optimization simulation category in Training Shop.
Our services are not limited to the mentioned subjects, and the MR-CFD is ready to undertake different and challenging projects in the Machine Learning/Optimization Engineering modeling field ordered by our customers. We even accept carrying out CFD simulation for any abstract or concept design you have in your mind to turn them into reality and even help you reach the best design for what you may have imagined. You can benefit from MR-CFD expert consultation for free, and then order your project to be simulated and trained.
By outsourcing your project to the MR-CFD as a CFD simulation freelancer, you will not only receive the related project’s resource files (Geometry, Mesh, Case & Data, …), but also you will be provided with an extensive tutorial video demonstrating how you can create the geometry, mesh, and define the needed settings(pre-processing, processing and post-processing) in the ANSYS Fluent software all by yourself. Additionally, post-technical support is available to clarify issues and ambiguities.