Please help us help you and fill where relevant:
Your LimeSurvey version:   LimeSurvey Community Edition    Version 5.6.7+230222
Own server or LimeSurvey hosting: Own server
Survey theme/template: fruity
==================
Hi everyone,

I generated 5 random numbers 1 to 5 using the following code, to create three different groups.
If random number 1 and 2, treatment group one. If random number 3 and 4, treatment group two. If random number 5, control group. 

Now I would like to show different information for these three groups conditional on the randomised numbers, which also worked out. 

After presenting the information, in the next page, I also want to check if they understand the information correctly. Therefore, I created other questions conditional different groups. And if their answer is wrong, they go to another page to read similar information again. These steps, I set the conditions on the randomised numbers, it won't work anymore. All the order of the questions and questions between different groups got messed up. The questions in treatment group one also got showed up in treatment group two, and so on. 

My colleagues told me that the limesurvey might not save these randomised numbers as fixed value, but they roll the dice every time. That is why it ruined the orders of the questions and the information between groups. I am wondering if we could save these randomised numbers as fixed value somewhere. So it does not ruin the orders of the question and the information between different groups. Or there are other better solutions?

Thank you very much for your kind help in advance.

My colleagues told me that the limesurvey might not save these randomised numbers as fixed value, but they roll the dice every time. That is why it ruined the orders of the questions and the information between groups. I am wondering if we could save these randomised numbers as fixed value somewhere.

If you would just create a random number, this would be actually true. However, this is prevented by using the construct you are using.

Bad would be: Because this way, whenever you call this equation question the dice would be rolled again.

However, by checking if the question is empty and if so roll the dice, if not use the value of the question, you have basically "fixed" the random number, once it has been set once.
I am currently wondering if .NAOK makes sense here. Because per definition, there must be an answer to randExp if it is not empty. Not sure if this could cause your issue, but then you say that parts run totally fine.

I think there must be something wrong in the rest of your survey structure, so I suggest yo uprovide a LSS file that shows the structure with the issues.

Thanks a lot for your quick reply. No matter if I use .NAOK or not, it still did not work out. 

I uploaded my LSS file here. It is highly appreciated if you could check this out. 

In my limesurvey code, I condition on the random number 1, 2, 3, 4 separately, even if random number 1 & 2 are treatment group one, and random number 1 & 2 are treatment group two, because at the beginning I thought "OR" condition was the reason, but it turned out that it was also not the reason.

Thank you so much for your help!!!