I have to confess a certain bias here: https://www.researchgate.net/project/Vspice?_sg=9yzIQFPDSDp7CiAhqpvKvZ6cZid5yYX9Ez9klI5HvmeOSV_RN0ldIqxvabM8efIfxftNJiThOPs11BUihRpwmxjyN-JEGnI9MR-l

I would suggest that your editor was probably brought up on HTML and stuff like that, and didn't understand what he read...

1- Dont think too much about article review results. Sometime its up to your chance and the reviewer/s is/are assigned to your paper.

2- If you didn't provide enough evidence from your simulation from your experimental work or other researchers, actually second journal made mistake to accept paper. as a scientist you have to be able to validate/correlate your analysis with experimental measurements. Otherwise you are just an operator of an scientific software or just a code writer.

3- You've performed your simulation base on some assumptions and simplifications. More than your results, researchers want to know how these assumptions and simplifications can affect final results, then need to evaluate your simulation with some experimental evidences.

4- PLEASE NOTE: you have to provide scientific/experimental evidences for you simulation results not really do experiment.

Dear Amir Reza Ansari Dezfoli sir,

Our results were in good agreement with related experimental works. Also, we compared ours with literature. Thanks.

Simulation using a computer has the effect of adding to the theory on which the program is based, with some actual results. These are limited by the program, but they do allow us to compare them with the experimental results. When the comparison is close it can help to confirm the theory. By this means we can better understand the way the phenomena is organized in terms of cause and effect.

They can be a roadmap to future experiments based on both expected and unexpected results.

I agree with Rick Manner answer.

Hi Have a look at the following article

Computer Simulations in Science

https://plato.stanford.edu/entries/simulations-science/

In several research projects together with industry - particularly with divisions within the VOLVO sphere, like VOLVO Aero (now GKN Aerospace), VOLVO Penta, and VOLVO Trucks, we combine simulations with a mathematical optimization model created by ourselves, in order to optimize designs under physical constraints. We are getting very good results, and several PhDs have received their doctorates through these collaborations.

Simulations are interesting.

In chemical sciences simulations or modelling are often published.

What I expect is to prove at the end that the "model" "fits to the purpose", i.e. some validation must be performed and it must be proved that we are able to obtain results by simulation that fit into the real measurement performed for same conditions as for the "simulatation".

regards,

GB

my standpoints are that the simulation could help me find some new problems that I can not get by imaginations or by logically thinking, and even simulations enable me to get some new inspired ideas. of course, the simulation environment is far from practical setting.

Simulation are very valuable to companies that operate commercial units when they need to make decision about alternate feeds, products, operating modes and turnaround and maintenance schedule.

In my research I study the behaviour of production systems. The existing literature used discrete event simulations heavily to understand and analyse the behaviour of the production system. However, I see that those simulation models are far from reality. Even production and maintenance managers of the manufacturing companies (associated in my research project) of the production system sometimes don't trust the simulation results. This is because they are built on vast number of assumptions and simplifications. For example, in production systems production and maintenance personell carry out improvement activities continuously and also implement some structural changes. That means the inputs to the simulation models needs to be updated continuously. Analysis based on the static inputs becomes less accurate to be able to make decisions as soon as improvements are made. Moreover, building simulation models takes long time and efforts. And to keep the model updated we need to have expert coders and this skill is often difficult to get ( that increases the waiting time to make new decisions) and more expensive. These are some of the practical limitations of simulation in production systems. To be able to provide a better decision support to production and manitenamce managers, companies are increasingly focusing on analysis of big data sets of machines ( the consequence of digitalisation is data deluge) to be able to provide the same decision support as simulation for practical purposes.

That being said, simulation is a great tool when they don't have any data collected from production systems and verify and validate new scientific theories and ideas, pushing the bourdaries of existing research. But those theories and ideas needs to developed further and be adapted to the real-world scenario in order to able to use them in practice.

To summarise i think the researchers can use the theories proved in a simulation environment and develop it further by conducting experiments using the real-world data and including the practical aspects and trade-offs. I think this will help the industrial community to better be able to understand and use the scientific outcomes.

Md. Mahabub Alam Moon :

It's not clear, but the question seems to imply that simulations model the ideal case exclusively. It is true that a simulation is theoretical - however, if the model does not account for non-ideal behavior (assuming such behavior exists in the "real-world" event/process/etc. being simulated) then this would be considered an incomplete or arguably "bad" model and only if the impact in not factoring in such behavior is significant to the result.

I just wanted to be sure that there was not a false equivalency here between models/simulations and "ideal".

One dimension of the utility of simulations is that some simulated events cannot be conducted experimentally. Not directly and not in real-time. Some events may take longer than our lifetime to conduct an experiment whereas the model does not have to be "computed" in real-time. A solution set can be arrived at on a shorter time-scale than the experiment. An experiment can validate some small time-slice subset of the simulation. But perhaps the experiment is not feasible due to temporal requirements.

I understand that the specific domain which is pertinent to the question is reasonably well understood and predictable in gross terms. The work I hAve been associated with has used simulation to compare different says of managing anthropogenic effects on a system (in our case, coastal marine systems with marine industry). It isn't possible to do such studies in the real environment -- the each strategy trialled would alter the baseline for all subsequent strategies. In a simulation, it is also possible to play "what if" games with the subject ... possibly games which would be much more expensive in vivo.

It's horses for courses (of course), but simulations of systems ad.its a great deal of exploratory work comparatively cheaply ... at least in the domain I am most familiar with.

Towards the end of the Cold War, I was running simulations of the effect of EMP on electronic devices. I would suggest that the simulation was a better approach than detonating a nuclear device over Sunbury-on-Thames....

A wise man once said that "No simulation is perfect but most simulations are useful. "

I have quite many textbooks written by scholars in my office shelves - more than 1000. Rather many textbooks in convex analysis, game theory, and related fields from - say - the 50s to the end of the 80s have plenty of cases from the cold war: min-max (saddle-point) problems illustrating strikes and counter-strikes with robots between the US and the Soviet Union, resource allocation problems related to submarine detection in the Atlantic ocean, and the allocation of radar installations. Also I am glad that those convex-concave functions were never needed "live" - but only were simulated! But the PhD courses were a lot of fun, although I do not think that I will ever be in need of using the bordered Hessian!

Take, for example, Theoretical Physics: Often, the aim is to solve specific models

exactly. Unfortunately, only for a few cases, such exact solutions have been found. Indeed, usually, one may rely on various approximations, with an unknown accuracy. Here, simulations, especially, Monte Carlo simulations or Molecular Dynamics, play an important role, because they provide a systematic way to improve the accuracy.

More details may be found in standard books, e.g., by Kurt Binder, David Landau, and others.

Comparison of model calculations with experimental data is another aspect, which needs great care, simply, because of possible shortcomings of the model or uncertainties when performing the experiment.

In medicine, simulation has greatly contributed economically by proffering near-exact answers to human problems without breaking huge amount of money. It has shortened the experimental time from initiation to finalization. It has helped focus far forward than human experimentation with great outcomes for medical advancement.

Simulations are extremely useful when experimental studies could turn out to be cost prohibitive or it is near impossible to conduct experiments and/or collect data. Modeling via simulations should be done with simpler systems first so that they can be validated with experiments. Then the models may be extrapolated to larger and more complex systems so that the effect of independent variables on one or more dependent variables can be simulated (predicted). See for example

Conference Paper Simulation of Plasma Reactors: Etch Rate of Polymers

Monte Carlo simulations and finite elements modeling have been used for decades. Computational Fluid Dynamics (CFD) is finding more an more applications in modeling vascular flow i human bodies to pinpoint potential areas where cancer could develop.

Srini Vasan My last experience with Monte Carlo analysis was some software I wrote for a gaming company, to predict the winner of a horse race. It was based on form of both horse and jockey over tracks and distances. The software had a hit rate of 75% and everyone in the office used it to bet on the Melbourne Cup. Smallest win was $2700 and the largest was $30,00 - which absolutely proves the usefulness of simulation...

Concerning the second question,

To perform real tests, one needs access to expensive hardware and skilled personnel. Additionally, real tests usually consume a considerable amount of time and money, and it may be hard to reproduce the same scenario several times.

In this context, the development of simulation frameworks that allow transferring real world scenarios into executable models are highly relevant.

However, simulation frameworks have their own drawbacks, as it is impossible to fully reproduce the real environment.

Simulations are useful for many reasons. Originally, they were developed to perform tests that are impossible, extremely difficult, very expensive, dangerous, too resources demanding, harmful to people and nature, and many similar.

Think about the testing of concrete containment of a nuclear powerplant, airplane crashes, car crashes, spacecraft approach to a planet, safety tests, etc. Are you going to involve in those tests real, usually very expensive devices/infrastructures, or do you prefer to test them virtually?

Think about testing new safety features on a car that shows its effect only at high-speed crashes. How many real cars do you want to crash until you find the ideal design? In the computer, you can simulate hundreds or even thousands of different shapes of the given car parts until you find the ideal one.

Are you going to crash 1000 cars just to improve one simple part of a car? Think, in the car are thousands of parts. Are you going to taste them all in real cars?

With research, it is the same. Instead of performing thousands of real experiments, we simulate and develop our understanding of the subject according to already known experimental evidence. Using simulations we extend our knowledge, which gets tested in just one experiment and not in thousands of them.

Usually, we have some preliminary experimental evidence that gets theoretically explained and extended. We build a model that gets simulated. The outputs of those simulations are compared with the new experiment(s) what in turn stimulates a new wave of theoretical models and simulations that got again compared with the experiment. It is a never-ending cycle of development.

Dear Mahabud,

In photovoltaic domaine, device simulation helps photovoltaics (PV) researchers to understand solar cells, to perform loss analyses, to predict how changes to a solar cell design or fabrication procedure will affect its efficiency .

Good luck.

I can add that a simulation is a necessary step forward before real tests. In general, researchers think that the best experimentation includes 4 consecutive steps: Simulation, mixed reality, controlled environment, and open world.

I. About 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...

There are some kinds of research where a direct experiment is not possible because its results are unsafe or damaging. For these situations a simulation can show the results without causing anybody any harm. For example in economics a change in the taxation regime would be politically damaging to a government which introduced it without first demonstrating by simulation how its effects would be felt. Unfortunately very few satisfactory simulations for representing our social system are available, and which work accurately because there are too many unknown factors at work. My 310 page e-book "Consequential Macroeconomics" and short article SSRN 2865571 "Einstein's Criterion Applied to Logical Macroeconomics Modelling" show the way but need more development to be effective.