These approaches complement one another. Without the theoretical simulations experimentalists might not know what to test for. Without experimental results the new theoretical models might have too many free parameters to be useful. So these methods go hand in hand, they both are necessary for scientific progress.

Dear Zakaria,

We can not split the experiments with the theory or simulations. Infact theories are built on the experimental evidences. So if we want to do a very good piece of scientific work, it should be backed by both experiments and theory.

In order to effectively address and tackle global challenges like climate change, food, nutrition and energy security, unemployment, etc. greater emphasis on applied research is the need of hour. However, importance of basic research research for generation of new knowledge can't be overlooked. As such, a sensible balance of both basic and applied research is crucial.

If scientific progress is equated with progress in the accumulation of scientific knowledge, then a theory that is not confirmed by experiment will not ultimately receive recognition in the scientific community. So what happened before: a chicken or an egg?

To my great regret, along with the amazing achievements of mankind, there is a tendency when you come to the conclusion that over the past 20 years the lion's share of scientific and technical articles cannot be refuted or confirmed. Original ideas, designs of various equipment, various technological processes developed by talented scientists and engineers are presented by models in scientific works without comparing the magnitude of the experimental uncertainty achieved, with the difference between the theoretical curves and the data of tests or experiments. It would seem that this requirement is the basis for the advancement of theoretical thought in practice. However, for example, only 20% of International Journal of Refrigeration articles meet this requirement. To paraphrase Lenin’s famous quote, how far away are all these models from people. The percentage of talented articles is decreasing. So what has a greater impact on scientific progress: theory, experiment or the desire to increase your citation?

Thank you all for very important replies. Actually Boris wrote very important words about the need of realization of the simulation research. I know many people who used just some software packages and use them to simulate chemical processes. The results are published in good journals bu I have not seen any one experiment to compare or realize the predicted results.

I would like to contribute to the discussion with some comments about the interrelation between theoretical and experimental work in the assessment of a new idea in process engineering, from the point of view of small companies that want to innovate but are located in poor countries.

The design, construction and operation of a pilot unit with the purpose of evaluating the possibilities of a new industrial process, or of a significant modification of a known process, can be quite relatively costly.

If the verification is really successful, nothing can be better than an experimental verification by means of a unit that materializes the process in question.

But if experimental research shows that the proposed idea is not viable, a considerable set of resources will have been used, generally in vain.

On the other hand, an idea that seems interesting can ultimately result in a successful innovation, crystallize in a new patent and produce significant benefits, for and through the company that risked putting it into practice.

What to do then, if you are not in a position to invest a considerable amount of resources to lost funds, as it may be necessary to invest in the research and development of a new process?

The use of digital computers has become widespread for several decades, its cost has significantly decreased in relation to the benefits that can be obtained from them, numerical methods and software packages that implement them have been developed, and the number of people trained or able to be trained relatively quickly to apply computation and digital simulation techniques in activities related to research and development.

However, techniques based on digital simulation are only applicable in engineering when the laws governing the process in question are well known and can be formulated at a precise quantitative level that allows the construction of mathematical models of the system as well as models for simulation, able to faithfully reproduce the material process considered. This is generally possible when it comes to physical or physicochemical processes, but it may not be possible in some cases involving, for example, biotechnological processes. In the latter cases, one cannot expect to obtain reliable simulation results, both when it comes to discarding and when it comes to confirming a new idea and the risk of having to discard the idea after experimenting with a pilot unit should generally be assumed input.

When they can be applied independently, in a first stage of evaluation of a new industrial process, the techniques of mathematical modeling and digital simulation of industrial processes constitute a valid alternative to evaluate the possibilities of a new idea at a cost significantly lower than generally involved in an experimental approach from the beginning.

The application of relatively low-cost computing techniques becomes especially important when the idea explored is ultimately not viable.

In a first stage the emphasis is placed on building a simplified model, such that if from it it is obtained that the proposed innovation is not viable, then a more sophisticated and more realistic model will surely lead to the same conclusion. In case the idea should not be discarded in this first stage, it will be possible to move on to a second stage of digital modeling and simulation, in which the effort will be focused on studying the influence of the phenomena that were neglected in the first stage.

If the apparent technical feasibility of the proposed process follows from the second simulation stage, this must still be confirmed by a pilot unit. But generally the design of the test experiments is greatly facilitated by the information obtained in the computer stages of the evaluation, its cost decreases significantly and the likelihood of the final success increases by decreasing the risk of the corresponding investment.

Without a doubt, Robert's remark is very relevant. However, the idea of ​​combining analytical and numerical methods in a computational experiment has been known for more than 40 years. So, realized in the form of a computer program, the mathematical model is a kind of computational experimental unit. At the same time, in the scientific community, the prevailing opinion is that the use of supercomputers, large simulations, and large-scale models can provide a high degree of approximation of the model to the studied object. For example, the standard Energyplus input file, developed by DOE (USA) as a beta test of the whole building modeling mechanism to describe the building, contains about 3,000 input data. Its preliminary calculated accuracy (uncertainty, for example, room temperature) is very difficult to evaluate, since it strongly depends on the accuracy of the input modeling data. Without measured data for comparison and calibration, the results of energy modeling can easily account for 50-200% of the actual energy consumption in a building. Therefore, it is not possible to verify the model and its results, but one can only increase the level of trust in them [12].

However, human intuition and life experience provide a simple truth. For a small number of variables, the modeler gets a very rough picture of the process being studied. In turn, a large number of accounted variables allows one to deeply and thoroughly understand the structure of the phenomenon. At the same time, with the apparent attractiveness, each variable brings its own uncertainty to the integral (theoretical or experimental) uncertainty of the model or experiment. In addition, the complexity and cost of computer simulations and field tests increase enormously. Therefore, an optimal or rational number of variables specific must be applied to each of studied process [ http://www.ejaet.com/PDF/5-9/EJAET-5-9-740-744].

Both theoretical simulations and real lab experiments contribute to each other and to the scientific knowledge. However, in my personal opinion, whether it's a theoretical simulation or a lab experiment-based, a research has a higher impact if it has some form of validation of its results.

Pure theoretical simulation or involving real laboratory experiments

In my opinion, it depends on the challenges being tackled. If for example, one is considering traffic control and real time applications, then simulation is important else lab work is good