autt

02-02-2007, 03:05 PM

Hi everybody,

I have worked for optimization for a long time, but about optimizing refrigeration system, looks like it is seldom applied, maybe optimization is more complex, there is no mature and reliable optimizing method for systems like refrigeration, or other reasons as whole range search usually take unacceptable long time.

The optimization includes many aspects as factory cost, running cost, thermal performance, control performance, technology, structure, media as so on.

I should say not all of the aspects can be calculated, it is unrealistic to perform all optimizations in one step, the optimization here to talk is just the factory cost and running cost, thermal performance is used as calculating input. The other part of optimizing can be performed in other step, by experience or calculating.

I don't know other people how to do the optimization, here is my idea about it.

First picture is the pressure balance diagram, it shows how to get fan head by using upstream and downstream pressure calculating method. The fan head is used for fan selecting, and cost calculation.

The second picture is a basic single stage refrigeration system optimizing diagram.

Optimize object is sumed from each unit, used to evaluate system optimizing effect.

Optimize quantities is the selected parameter for optimize process to adjust. The optimize process modify these parameters slightly and do the system design calculation to get modification effect, and then modify parameters again and again by state of effect. These quantities have data range respectively to avoid unreasonably trying value.

Constrain quantities are usually the design requirements, some of them can just be inputed; the other are calculation output patameter, they can be selected together with optimize quantities, the optimizing process will check their data range after system design calculation finished, to estimate if the modification make sense.

Discrete quantities are usually the integer value, the optimizing method doesn't accept these quantities, so they are modified by hand after one step of optimizing calculation finished.

Anybody has ideas?

I have worked for optimization for a long time, but about optimizing refrigeration system, looks like it is seldom applied, maybe optimization is more complex, there is no mature and reliable optimizing method for systems like refrigeration, or other reasons as whole range search usually take unacceptable long time.

The optimization includes many aspects as factory cost, running cost, thermal performance, control performance, technology, structure, media as so on.

I should say not all of the aspects can be calculated, it is unrealistic to perform all optimizations in one step, the optimization here to talk is just the factory cost and running cost, thermal performance is used as calculating input. The other part of optimizing can be performed in other step, by experience or calculating.

I don't know other people how to do the optimization, here is my idea about it.

First picture is the pressure balance diagram, it shows how to get fan head by using upstream and downstream pressure calculating method. The fan head is used for fan selecting, and cost calculation.

The second picture is a basic single stage refrigeration system optimizing diagram.

Optimize object is sumed from each unit, used to evaluate system optimizing effect.

Optimize quantities is the selected parameter for optimize process to adjust. The optimize process modify these parameters slightly and do the system design calculation to get modification effect, and then modify parameters again and again by state of effect. These quantities have data range respectively to avoid unreasonably trying value.

Constrain quantities are usually the design requirements, some of them can just be inputed; the other are calculation output patameter, they can be selected together with optimize quantities, the optimizing process will check their data range after system design calculation finished, to estimate if the modification make sense.

Discrete quantities are usually the integer value, the optimizing method doesn't accept these quantities, so they are modified by hand after one step of optimizing calculation finished.

Anybody has ideas?