Dynamic models should always be for "fit for purpose". Dynamic simulation opens up the ability to model the real world with a bit more reality. The key is to understand that having the ability to model something with more detail doesn't mean that it is good to do so. Every attempt to add detail and complexity to a model has to be evaluated against the expected return. This argument also applies to plant piping. The purpose of the model is the key to selecting the modelling approach.
The key parameters for selection are: pressure drop, lag (holdup), dead time, phase slip
1 - In many cases we need to capture pressure drop. Do we have a reference pressure drop/flow that can be used for a quick CV estimate? If not, Do we have an equivalent length (and size)? If not, Do we have detailed pipe geometry? Depending on the answer to these questions you try to select the simplest approach first, i.e. a valve with a CV is preferred to a pipe with detailed geometry because it is simpler (assuming it achieves the desired goals).
2 - Do we need to capture lag? There are many reasons to capture lag, or holdup in the system, and it is the second most common parameter needed in dynamic models. Adding volume to any of the simple methodologies is quite simple as described in Stuart's comment above. When DP is not of interest, and it is only the system lag/holdup, you see separators being used instead of valves with volume. Separators are more challenging though when multiple phases exists because they behave as if the system is not turbulent and every phase can separate, which is typically contrary to pipeline flows.
3 - Do we need to capture dead time? This is the key parameter for starting to add complexity in my experience (and the most difficult). Dead time can only be captured with a segmented approach (because HYSYS will flash a mixture within the segment), and even then you are always balancing the true dead time with the number of segments you add in your model to stay close to reality. You almost always need some methodology for estimating the expected dead time, then tuning your setup to achieve that. Complex but doable.
4 - Do you need to model phase slip? This is really where you should start thinking about a full blown dynamic pipeline application, rather than HYSYS and its standard components. That doesn't mean you need to abandon HYSYS though, there is a special module called Aspen Hydraulics that is available to most users on a token system. Usually, when dynamic modelling of multi phase flow is mentionned people immediately think of OLGA it would be good to have some competition in that market and having active users of Aspen Hydraulics is one way to achieve this.I am even convinced that SPT Group would benefit from this as it will make the modelling of multiphase pipelines more wide spread and even if they loose some marketshare, total usage will go up.