Introduction
Everyone thinks the simulation, regardless of which it is, will solve the problems and bring the necessary answers to the doubts. However, the simulation is not a kind of magic or oracle that has ready answers for our needs, it must be properly built to bring satisfactory results.
But after all, what is simulation?
Simulation is the representation, imitation of a real system, in other words, through models we try to represent a particular system, in order to study it and get some results.
I usually say that the results are a reflection of the created simulation model, if we build a model that does not represent properly the system, the results will bring no benefits, or worst, may lead to wrong decisions.
The model must be constructed looking to be as close as possible to the real system under study, for that, some steps should be taken to obtain better results consistent with the expected goals.
The 7 steps
For proper performance of discrete event simulation, there are some steps that should be followed to achieve success with the study. Follows the seven steps that I believe are necessary:
1 - Determine the goals
Setting the goal is the first step to be taken. We should never start a simulation without having a purpose to be achieved. The goal should be formulated clearly, highlighting the issues for which we desire to get answers. The discrete event simulation can be used for a multiple purposes, among the most common are:
- increasing productivity in existing production systems;
- assist in decision-making in new facilities investments;
- sizing process inventories;
- analysis of material flow;
- sizing manpower;
- continuous improvement of the production process.
Importantly: with more goals more complexity to build the model
2 - Perform an appropriate data collection
In this step we must analyze the actual system and check what information is relevant to construct the model. It’s important to check which are the input datas and what should be expected from the output datas to allow build the information that will serve as a response to the goal. The data collection process is the most time consuming phase of discrete event simulation, we must take time to understand the process and to collect data, thereby allowing the activity occur properly and to build the model as close to reality as possible. For the success of this step, the process must be carefully analyzed with the goal to obtain the largest possible amount of data. In case of new production systems, we should seek references in similar process, to make possible to collect data that will help to the proper construction of the new model.
3 - Build the model
Through the use of dedicated software for this type of simulation, we should compare the tool objects and the elements that make part of the process. Thus, it is possible to construct a model that represents the system under consideration. This step requires a certain time, it is necessary to create a model that can adequately represent the system that is in study, the logical and procedures should be created to allow the model be able to represent the real system and achieve expected goal.
4 - Validate the built model
With the model already built it's necessary to validate the data that were considered for their construction, in other words, we should check if the model is actually working as real system in case of existing processes, or in case of processes not yet available we should see if it's behaving as imagined. This step is important because if there is no data validation, simulation results can be compromised.
5 - Perform simulation and collect the results
After the data has been validated, we must start the simulation and collect the results. For significant data collection, the model must remain running for a certain period of time, thus it is possible to obtain more reliable results. When in doubt, we should run the simulation as many times as we feel necessary, after all that's why we are simulating. At this stage it’s still possible to make some changes in the model to see how it behaves and evaluate other possibilities. The most important thing is not to make huge changes in the model, but changing only some parameters and see how the model behaves, otherwise you will be falling again in step 3.
6 - Analyze the results
In this step we need to critically analyze the results and turn them into information to aid in decision-making and serve as answers to questions that were part of the goal. Through this analysis we can check which way to go as well what should be done to the process in question.
7 - Make the final documentation
As a final step, the documentation should be prepared with the information obtained from the simulation, describing in detail what should be done. This document is important to specify and detail the necessary paths to be followed for the changes that we will do.