The design of units is a major part of the game and there is a large semi-random part to it.

While you can run a long time with the initial designs - particular if you selected a higher tech start - a design council should still be early on your list.
Latest to work in new Technology. But even before that, a design pass or new design line can give you a massive military powerspike.

Not a official Term, but I use those two to differentiate between 2 very different uses of the Model Design Council.
Design lines start with a single model that you make from scratch.
New Design Lines are not that common, but can arguably be more important then mere passes. New Design lines are the only way to increase the Structural Design - without a doubt the most import and static stat of a Design Line

Afterwards, you can only make passes over this design line.
Design passes are very common. If you want to translate Field testing into better stats or work in new Armor and Weapons or Linear techs, they are the primary way.

Strucutral Design:
As the value is between 70 and 130, the average would be 100.

Original Design:
"The Structural Design Score is randomized with the first version of a Model and will not change with subsequent versions." - 5.12.3.6

Subsequent Designs:
This value is fixed, for a model line. It will not change.
If you want a higher value, you have to start a new line from scratch, meaning you are rolling a new base design and do new field testing. A low score it is the one reason to start another line.

Base Design:
Also sometimes called "Field Testing". All other important values are based on it (and Structural Design).

Original Design:
The Value is Randomized between 70 and 100.
While a high value is better, if it is fielded any "kinks" can be worked out.
But even the best design is not so good, that field experience can not make it better.

Subsequent Design:
The Previous Designs Base Design + Field testing done with the design. This value will only go up until it will hit 130 eventually, but the progression becomes exponentially harder the higher the value becomes.

Weapon/Armor/Engine Design
The formula for all 3 is:
(Base Design / 2) + 4D20, then "modified by structural design" - http://www.matrixgames.com/forums/fb.asp?m=4830993
Note that the large amount dice used, will result in a very strong trend towards the average of (Base Design / 2) + 42. The more dice you roll, the less likely extremely high or low results are.

Original Design:
These values will be rolled once, then used. The values before structural design can be as low as 35+4 = 39 (minimum base design and roll) or as high as 50+80 = 130 (highest base design and roll). With a average around 85/2+42 = 42.5 + 42 = 84.5 - so 84/85 equally likely.

Subsequent Design:
These values will be rolled based on the (propably improoved) base design. They are then compared to the previous design values - and wichever one is higher will be used.
So they will never decrease, but are not guaranteed to increase either - if you roll lower, it just stays by the previous design values.
At maximum base design of 130, these values can go up to 130/2 +80 = 75+80 = 155, before structural design. But such a high result is extremly imporable. 75+42 = 117 seems to be something you can expect to get.

Weapon Design:
This value is applied to the Firepower of the weapon. Wich is turned into the weapons Soft Attack and Hard Attack values.
If you want real life examples for bad weapons designs, see the M16 Rifle and the Mark 14 Torpedo - but both were also worked out by proper (field) testing.

It does not seem to have any effect on the Calliber/Armor calculations.

Armor Design:
This value is applied to the Units Hitpoints, wich is derived from the base unit and the Armor equipped.

It does not seem to have any effect on the Calliber/Armor calculations.

Engine Design:
This is applied to the engines Carry Weight. Wich is then used against the fixed weight of all gear (engine included) to get the mobility Modifier (5.12.3.7).

Just because you mandated a strong engine, does not mean the designers can not still mess up:
Gears, Transmission, Cooling, weight distribution, Fuel/Electrical Lines or Fuel Tank/Battery sizes on the unit.
And that is before we come to the point where the production model could be a good 10-20% heavier then the design goal had anticipated!
In the game, all this seems to fall under "Engine Design".

Field Testing Redesign:
After some amount of field testing, just doing a design pass on the same gear is worth considering.
Even if there are no semi-passive techs to work into.
With the higher base design, there is a decent chance you will get better rolls for Armor, Weapon or Engine design.

Semi Passive Tech Redesign:
Armor skirts
Improved Sensors
All those linear optimsation techs

These techs are semi-passive bonuses: They do not seem to cost any material or design time to install. However they can only be included if you make a design pass.
They will automatically be grabbed and added when you make a pass anyway, but you can make a dedicated pass just to work in those new developments.

Empty Redesign:
If you feel the Weapon, Armor or Engine Design rolls are too low for the Base and Structural design, you could just make another design pass without even deploying it.

Changing Gear:
Especially early on, you will often change the gear. Infantry with Automatic Rifles and Combat armor are usually a good early goal.
Depite what I and others said previously, there is currently no indication that changing gear has any impact of field testing.
Apparently experiences with a Carbine apply to a Automatic Rifle or Laser equally.

Practical Example for new design:
With a structural design of 87, the Grunt was pretty low. Not really worth developing further. My chances to roll better were pretty good.
So I made a new design from scratch, wich came out as the Soldier-at-Arms. The only change I made, was upgrading the Armor from Enviro Suit to Padded Enviro Suit (+50 to +100 HP)

Strucutral Design was a new roll, that ended up being above 100 and a good 20 better then the Grunt.
Base Design for the Grunt 2 would have been 80+15 Field Testing=95.
And SaA's 91 BD score still is acceptable compared to that. It will keep going up the more I field it.
Weapon Design made a jump of 16. So despite the weapon staying the same, they can use it a lot better. Also note that with a Firepower of 100, the Soft Defense value seems to match exatly the WD score.
Armor Design had a slight decrease from 91 to 90, but with the better armor that still more total HP. Also note how HP with a Armour Strenght of 100 eactly equals Armor Design.
The Grunt 2 could not have gotten lower then 91 (and might have rolled higher then that), but the value is comparable enough. Meanwhile the long term prospects for the "Soldier at Arms" design are much better. This better Structural Design will pay dividends every future design pass.

Now that you have the new gear, it is time to actually deploy it. Unfortunately it is very easy to forget some units in a forgotten formation. Here is a way to avoid that, by using the Replacement System as your ally.

0. Do not set a Model to obsolete, unless you got 0 instances left or really know what you are doing.
Obsolete is treated as Disallow for every last unit on the map. Gets subunits send back to the SHQ, even if that ends up completely vacating units. People lost the games doing that accidentally.

1. Set up all units to tollerate Low, but keep Regular and higher just on "allowed". This only has to be done only once, but also once for each unit/formation.

2. Set the old model to Low Quality, New Model to Regular. You can lower the old model to Obsolete once you hit 0 remaining units

3. Put some Replacements of the new version into the SHQ, even if you do not have any losses. Think of the as reserves to be trained up.

4. Wait a turn and the Replacement step will happen:
As the old model is low quality and low quality is only tollerated, the SHQ will try to replace the old models with the new model in all field units.
As of 1.08.01 the game will automatically replace the gear in field units, so the XP stays in the front unit

5. You should now have some examples of the old model in your SHQ.
- If the old one can only be replaced and not upgraded, scrap them and add more Replacements of the new model. If they are not taken, check if that was really all (or they just ran out of range).
- If the model can be upgraded, upgrade it. They are now of the new, prefered model and will be handed out next turn - allowing you to slowly but steadily replace all the fielded units with minimal disruption.
When upgrading you can even consider "Upgrade all units under the same SHQ". Any unit you can not reach this turn might send it's units back next turn (for the new replacements), wich would be a reminder that there are still upgrades left to do