First understand some definitions:
Newtonian fluid: A fluid whose viscosity is not affected by shear rate applied to it. When shear changes, viscosity remains same.
Non-Newtonian Fluid: A fluid whose viscosity varies with shear rate. Viscosity does not stay same when shear changes.
Rheology: Study of fluid flow in non-Newtonian fluids.
All plastics are non-Newtonian fluids.
This means that their viscosity does not remain constant over a range of shear rates.
Strictly speaking, rheological behavior of plastics is a combination of non-Newtonian and Newtonian mechanics. At lower shear rates, plastics are non-Newtonian, but as shear rates increase, plastics tend to exhibit Newtonian behavior.
This is because as shear rate increases, polymer molecules begin to separate from each other and begin to align along flow direction. Reference is as follows:

 

In injection molding process, material is subjected to a large amount of shear forces during cavity filling stage.
Shear rate is proportional to injection speed. If shear rate is in non-Newtonian region of curve, then small changes in shear rate will cause large shifts in viscosity.
This will make mold filling inconsistent and affect mass-produced parts. Therefore, it is important to look for Newtonian region of curve within this region and to set injection rate (hence shear rate).
Viscosity profile of any mold can be obtained using an injection molding machine. Shear rate affects viscosity much more than temperature. Therefore, as long as actual melt temperature is within recommended range, a similar viscosity profile will be obtained during molding.

When plastic enters cavity through runner, melt has a certain temperature, pressure and velocity. All three variables are time-dependent, which means that value of each variable will change for a short period of time until end of fill.
For example, melt temperature decreases with time. If a melt temperature of 280℃ is injected, after one second, melt temperature is lower than 280℃. Final size and quality of each injection molded product depends in part on temperature, pressure, and speed.
Consider a one-cavity mold:
Melt temperature at the end of filling was 450 degrees Fahrenheit, plastic pressure was 8000 psi, and speed of plastic into cavity was 4.5 inches per minute. Now, if temperature drops to 400 degrees Fahrenheit, part shrinks less and resulting part is now larger than before. Similarly, if end of fill pressure and velocity changes, size and/or end of part will change.
Now consider a two-cavity mold, each with same cavity size:
If two cavities are not filled with similar filling conditions, then from discussion above, we know that two parts produced from each cavity will be different.
This is why a cavity balance test is required.

As plastic flows through different parts of machine and mold, there is a loss of pressure applied at flow front of plastic due to drag and friction.
Also, as plastic hits walls of mold, it begins to cool, increasing viscosity of plastic and requiring extra pressure to push plastic.
Plastic skin that forms on walls reduces cross-sectional area of plastic flow, which also results in a pressure drop.
Molding machine has a limited maximum amount of pressure available to push screw at set injection speed. Pressure required to push screw at set injection speed must never exceed maximum available pressure. In this case, process becomes stress-limited.
During process development, knowing pressure loss for each section helps determine the overall pressure loss and pressure drop sections. Mold can then be modified to reduce this pressure drop and achieve better consistent flow.

Injection of plastic into cavity can be divided into two main stages.
The first stage is injection stage. During injection molding stage, mold cavity is completely filled with molten plastic.
The second stage is feeding stage. Feeding stage follows injection stage. Packing pressure must fill mold cavity, and plastic equates to volume shrinkage that occurs as plastic cools, as plastic hits cold walls of mold.
In most cases, packing and holding phases are not distinguished and are collectively referred to as packing phase.
Ideal packing pressure is determined by evaluating process window of mold.
Processing window is also known as forming area map. This is area where acceptable parts are molded. The larger window, the larger range of allowable shaping fluctuations. as follows:

 

Process is set in the center of this window so that any changes within window will result in acceptable sections.

Plastic enters cavity through gate. Plastic can enter or leave cavity as long as gate does not freeze.
Therefore, packing pressure must be applied until gate is frozen.
Do a very simple test to determine dwell time:
Weigh samples with different holding times, as holding time increases, more and more plastic enters cavity to increase weight. However, once gate freezes, plastic cannot enter cavity and part weight remains same. This is called gate freeze time or gate seal time. See below:

 

In image above, part weight remains same after 9 seconds. Dwell time was set one second higher than gate seal time to ensure that gate was frozen during each pour. In the case of image below, time is set to 10 seconds. This will ensure consistency and any small changes will be compensated for.
Injection molding cooling
Once plastic touches walls of mold, it begins to cool.
Mold is closed until cooling time is over. Mold is then opened and part is ejected. Before mold opens, part must reach an acceptable ejection temperature for plastic.
If part is ejected before reaching acceptable ejection temperatures, part is too soft and will deform during ejection. excessive cooling time is just a waste of machine time and profit.
Determining right cooling time is complicated.
In sections with thick sections, it is difficult to measure internal temperature in the center of thickest section. In some parts of mold, it was difficult to get adequate cooling.
Cooldown changes also affect shrinkage.

 

In graph above, dimension A (blue) is not affected by cooing time frame test. However, dimension B (red) varies with cooling time. Target value for dimension B is 0.135. So we can set a cooldown around 17 seconds.