Optimization by System Simulation
This is also one of my favorite topics.
I don't think real optimization can be achieved after the plant is installed since the effects of the original design decisions are so overriding.
Here is a simple example
What Iceman says about the discharge pressure is quite right. So how do you do this? Make sure that the air cooled condenser is kept clear and that it is clean...
But you can only do so much with this. As Gary pointed out in another thread, "you can add as much condenser area as you want but the sub-cooled liquid can never go below ambient".
If, however, the designer chose to use a water cooled condenser then the discharge pressure would be much lower.:( So you can't really optimize this system without changing the condenser!
What exactly is optimization?
You may want to optimize the flow of chilled water to be supplied underground. In this case, I expect that you would be concerned with low chilled water temperature and chilled water flow and not plant power.
So optimization depends on what you are trying to achieve.
Usually, you would choose a target parameter (for example plant power usage) and you will select all of the variable parameters that you can control to minimize (or maximize) the target parameter.
In an ideal world, you then run some experiments where you can change each variable at will.
In the real world, however, you don't always get this freedom. Even if you identify the chilled water flow as a potential problem, you can't easily increase the flow rate without making system changes.
Satisficing solution
The next best thing is to find a suitable answer. Optimization curves have a convex shape and there will be a band where you can make big changes for small gains. The good old 80/20 rule.
The items listed by Iceman are good general practice for running plant but they are not optimization.
To do optimization, you need to keep good records, make systematic changes to the plant and monitor the result.
You make a proposition. What if we did... Then you make some changes to see if the proposition is true. Sometimes, you can make predictions by studying existing plant. A good option lately is simulation. With much less cost, you can predict the effect of changing a parameter any parameters.
In this way, you get an idea of what is realistic and know the sensitivity of the parameters. The most sensitive parameter is the one that will give you the best results.
Stoecker simulates a simple vapor compression cycle and shows the relative component sentitivities as:
Compressor = 6.3
Condenser = 1.3
Evaporator = 2.1
So the logic for this example is start looking at the compressor. An improvement here will be 5x more effective than work in the condenser.
I hope we get plenty response in this thread. I think it is a most valuable area of our work. Especially in light of our current world energy crisis.