Answer
Add a normalization step at the start of the workflow: use a script task to check which variable name is actually present in the execution context (execution.getVariable('OrderNo') vs the SQL-style name), and if the one your downstream logic expects isn't there, explicitly set it from whichever one is (for example execution.setVariable('ORDER_NO', execution.getVariable('OrderNo'))) before your main logic runs. This way the rest of the workflow can always reference a single, consistent variable name regardless of which trigger path supplied the data under a different naming convention. Event substitution fields (documented under Manage Event Actions) are worth checking as an alternative if you're triggering purely from an event rather than needing this kind of dual-entry-point normalization.