2022년 8월 17일 수요일

Plotting with Tensorflow

Finding a reaction rate is one of the most essential procedure in chemical reaction engineering. This enables us to estimate not only performance of the reaction, but design of the reactor, and eventually to scale-up the plant. 

Determining of reaction rate is not a simple work even with all kinds of reaction data in various reaction conditions in hands, and even though it follows a very good elementary reaction. The reason of this difficulty is simply because experimental data is actually a combination of many of physical and chemical phenomena, and we are only looking for a way to find a reaction rate out of the data of very complex phenomena involved.

Mathematical calculation with formula, numerical calculation, or graphical method of calculation often does not satisfy what we really want in terms of plotting data with relations of one of the reaction variables. We just cannot separate one phenomenon to another by looking at data itself. 

Classical approach of determining a reaction rate always requires certain assumptions; isothermal, isobaric, uniform concentration (or perfect mixing), no residence time distribution, type of reactors and many others. How can we possibly match these assumptions with actual data? 

Plotting between variables and reaction results is a great method to give an insight of reaction behavior. This plotting procedure I would say is the most important work in chemical reaction engineering. At the same time, this plot needs to have a certain tendency for engineers to understand or utilize the meaning of Lab experiments. 

To give a meaning of a trend in experimental data is a very hard work, and not to mention determining reaction rate. Well, we can simply forget all the conventional or classical approach of reaction engineering for now, and let the computer work to find a relation with numbers instead of formula.      

 



댓글 없음:

댓글 쓰기