DirkZ70 - The limit you are hitting is actually a HTTP URL limit since all pipelines are essentially executed by a REST request.
I'm curious about your statement: "we use pipeline parameters to transfer information from main pipeline to sub pipeline." Do you mean to state that you are passing your data payload via pipeline parameter?
Did you know that if you leave an open input in your child pipeline, the data payload from the parent will be sent directly to it, avoiding the overhead of passing the data as query parameters?
My recommendation is to redesign your pipeline to pass data as a document to an open input view in your child unless you have overriding reasons to pass certain data elements as pipeline parameters.
Also, unless you need each input document to be processed in a separate child pipeline, you may want to try to enable the "Reuse executions to process documents" or use a Batch Size greater than 1 to allow multiple input documents to flow to your child pipeline (as if it were part of the parent). This would also improve your performance by avoiding the pipeline initialize/close overhead. See my article on the Pipeline Execute snap for more details.
Hope this helps!
