This repository contains all necessary files to deploy the ICS simulator on Heroku. All necessary dependencies are contained in both the environment.yml and requirements.txt files. The author created the simulator in an Anaconda environment using Python 3.7.

ics_main.py is the main script containing the simulator. The script uses the Streamlit library for web app integration. No HTML or CSS is required to deploy. There are very few parameters one would ever need to adjust within ics_main.py, but each will be lited below.

These can be found in the main() function within ics_main.py, stored as a list in the variable passwords. The passwords are used to assign the number of tests to be performed. This is done in the perform_calcs() function, as a dictionary stored in the test_dict variable. The names of the passwords and the number of tests can easily be changed. Remember to change the password names in both locations.

The administrator password bypasses the built-in wait time that the other passwords have.

These are found in perform_calcs() and are the threshold values taken from the CDD.

The outputs of the simulation are based on creating a normal distribution for each KPP to determine how far from the CDD threshold value the KPP is allowed to deviate. These are the p1, p2, and p3 variables in the perform_calcs() function. They are generated from Numpy's random.normal function, which take the mean and standard deviation as arguments. A user selection of a "low" parameter, such as an engine size of "small" of a fuel capacity of "minimum" invokes p1 for that KPP, which has a mean of less than 1. Selection of "moderate" and "high" parameters result in p2 and p3 respectively, with means of 1.0 and greater than 1.0. The standard deviation for each random variable p1, p2, and p3 are generated by randomly sampling a uniform distribution. This sample is then multiplied by the p1, p2, and p3 means respectively, in order to create a standard deviation that is proportional to the mean. Numpy's random.uniform function takes a minimum and maximum value for the distribution as inputs.

A dictionary, param_dict, is used to store the random variables, and then the variable is matched to the user's choice. The KPP threshold value is multiplied by the random variable to create values that vary from test to test, simulating a real-life scenario where different prototypes and tests have different results. The random variables are re-created for each parameter for each test, so that no two results are the same. The speed parameter has an additional variable, form_mult which accounts for speed gains that can be achieved by using different hullforms.

This can be adjusted by changing test_len, which is set in the progress_bar() function. The first part of the if statement is for the administrator password, and that should be left ar 1 second for the purpose of testing code functionality. The second part sets the time for all other passwords. The test_len parameter is in seconds.

Note that as the user orders more tests, a learning curve is achieved, and every 5 tests the total time to conduct the next 5 batches of tests is reduced by test_len. That is, let's say the test_len parameter is currently set to 120 seconds (2 minutes). If the user orders 15 tests, the test time parameter computes the total time to run all 15 tests as 1440 seconds (24 minutes). Note that this is less than 30 minutes. Each batch of 5 tests reduces the time by 2 minutes. Therefore, 3 batches of 5 tests reduces the total time by 3 * 2 = 6 minutes.

When running, the flow of information through the script is as follows:

Heroku runs the script ics_main.py. The statement at the very bottom of the script

if __name__ == '__main__': main()

causes the function main() to be called. main() displays the title and asks the user for a password. The user input is converted to lowercase and compared to the list of passwords. If the password is invalid, the user is prompted to try again. If the password is in the list of passwords, get_selections() is called.

get_selections() creates the drop-down menus for the user to select the ship configuration. All of the conditional if statements prevent the other menus from appearing until the user makes the prior selection. Once the final selection is made, all selections are converted to lowercase, and selections_complete is converted to True. This renders the 'Begin Testing' button. Once the user clicks the button, perform_calcs() is called.

perform_calcs() initializes all variables, creates the dictionary matching the password to the number of tests, and loops through all tests, computing a new KPP values each time. It also caps the Ao value at 1.0. All computations are completed when perform_calcs() is called. However, to create the imposed wait time per test, progress_bar() is called after computations are complete. The last 3 lines of perform_calcs() call progress_bar() and then calls show_results() when progress_bar() finishes. This is done by having progress_bar() return the string 'complete'.

progress_bar() renders the progress bar on screen, advancing it based on how much time has elapsed. The loop will be executed once for each second of wait time. The call to time.sleep(1) forces the loop to wait for 1 second prior to executing the next iteration. This is what actually creates the wait time. Otherwise, the entire loop would be executed at machine clock speed, and the whole thing would be finished imperceptably fast. The administrator password sets test_len = 1, which allows the program to be executed in mere seconds, but is slow enough to allow the administrator to see that everything is functioning as expected.

Once progress_bar() is complete, it returns the string 'complete' back to perform_calcs(), which is stored in the variable prog. Once it sees that prog == 'complete', show_results() is called. The purpose of show_results() is to display the average of the test results on screen, along with rendering graphs of all individual tests and their means. When show_results() has completed executing, there are no more function calls, and the program has completed running.