Organizing project portfolios for optimal results

The way I have done this in real life is through "horizontal/vertical" grouping of projects based on multiple factors including

1. Project Portfolios per ownership; which in most cases follows the "vertical" organization chart path, but can also follow the "horizontal" service or functionality path.
2. Project Portfolios per level: which basically differentiates between "corporate level" and "department level" projects
3. Project portfolios per Business value: where strategic (long term/deep impact) projects are grouped separately from shorter-term/less-impact tactical projects
4. Projects Portfolios per business initiative: where projects (possibly coming from different places within the organization) are grouped together per business initiative, whether coming from corporate level or business line level
5. Project Portfolios per resource requirements: where projects using  the same resources can be grouped together to avoid resource contention and maximize resource utilization
6. Project Portfolios by client : where service providers/implementers/integrators can handle groups of projects for the same client as one group utilizing common interfaces, common team members and common logistics in dealing with the client
7. Project Portfolios by vendor: where large customers may group projects coming from the same vendor (service provider) as one group
8. Project Portfolios by $$$ revenues/profits: where projects are grouped according to the revenues/profits thus leading to "key/strategic" project portfolios and the others following the 80/20 rule
9. Project Portfolios by criticality: subject to "specific" and "agreed"definitions of criticality,  those so called "critical" projects may be grouped together and managed in a special way due to the criticality level.
10. Project Portfolios by inter-dependency:  where projects and/or groups of projects which have high-degree of interdependency may be grouped together and managed separately

What is worth noting from practical view point here,

• Take the above classifications with high degree of flexibility. Nothing is cast in concrete, and it is after all up to the decision maker(s) to accept/approve what is most suitable in their specifics.
• The potential for high-level of dynamics in the relations between the different groups and group projects, where you may find one project in 1, 2  or more project portfolios at the same time, and you see groups moving from one project portfolio to another from time to time.

This implies the need for "high competency levels" both technically and politically when dealing with such topic.