Once your dataset is created, you can refer to it in JCL exactly as you would any other dataset. It does not matter that it was created dynamically, as the dataset disposition is included when you create it. It gets processed exactly as though it were created with a JCL DD statement. I'm not aware that there is even an indication that it was created dynamically, once it has been created. It is no different from any other PS dataset.
If cataloged:
//SOMENAME DD DISP=SHR,DSN=LOG.DYYMMDD.THHMMSS.OUT
If not cataloged, catalog it, then see above.
If deleted when closed, don't delete it but catalog it, then see above.
Note: I have assumed that you are creating your dataset in one JOB and accessing it in others. If you are accessing it in the same JOB, take good note of Bruce Martin's answer. Your dataset will be "hidden", from the normal assessment of Disposition processing of the JOB when it is submitted because the dataset is only created after that point, when the JOB is actually running (if it gets as fair as running, it may fail immediately with a "JCL ERROR" without even getting close to running).
Personally I'd do it in separate JOBs, but some people think they are keeping things simple when they are not.