Hello Engineer Mohammad Bilal,
Thanks for this interesting question.
First of all, unless it is clearly mentioned in the contract that the contractor should use this specific delay analysis method, you can negotiate with the client to use the relevant method that best fits your project and your situation.
As per AACE Retrospective analyses: “Retrospective analyses are performed after the delay event has occurred and the impacts are known. The timing may be soon after the delay event but prior to the completion of the overall project, or after the completion of the entire project… In other words, even forward-looking analysis methods implemented retrospectively have the full benefit of hindsight at the option of the analyst.”
Regarding your question and as per the AACE recommendation above, you should select the window date after the delay event and before the project’s original completion date.
Your statement “Do we need to add complete fragnet at start of delay event ,in the impacted window or we add only the portion of the fragnet limiting to window finish date.” needs more clarification. The fragnet you use should contain a set of events that are near to each other in dates and connected somehow. If you have more events with different dates, then you need to create another fragnet . However, the impact of the fragnet has nothing to do with the Retrospective analyses.
For example, let us say your data date is 30 June 2023, you can add events that happened before this date if you would like to use the Retrospective analyses method. On the other hand, if you are expecting events to happen after 30 June 2023, then you can include them in another EOT claim and use the Prospective delay analysis method.
I hope my answer cleared your doubts.