PDA

View Full Version : system modelling



Lc_shi
05-10-2005, 01:57 AM
Hi sirs,
We design refrigeration system usually basing on the standard working condition and select components. It's difficult to predict the performance off the standard conditions for the system is dynamic. Is there any good software for refrigeration system modelling or can we do by ourselves? pls advise.

thx

LC

US Iceman
05-10-2005, 05:19 PM
Hi LC,

A very good question.

Most refrigeration systems are designed for the worse case operating conditions (usually summer). In some cases, the design is modified to allow operation in the winter (head pressure controls).

If the system has capacity control, this can add another layer of complexity.

I totally agree with you that the systems are dynamic rather than static. The static conditions only apply to the manufacturers performance data.

Here is one link that may prove useful for you.

http://www.et.web.mek.dtu.dk/Coolpack/UK/

This was developed at the University of Denmark.

The main program used to create this solution method is a separate program called EES (Engineering Equation Solver).

You can find more information about this program at:

http://www.fchart.com/

This is a very powerful piece of software. I use it daily in my work. The software was written by two professors at the University of Wisconsin (Madison, Wisconsin, USA). It is expensive (about $1K US), but it is well worth the money. It has refrigerant and other fluid properties included.

It also has math functions that can be used within the program you write.

You can download a trial offer and see for yourself.

The EES program allows you to write your own programs, but be prepared to spend a lot of time doing this. However, once you are done developing the program, you have a very useful tool.

It does support different unit systems, but I believe all of the text and user manual is in English only.

Hope this helps you.

Regards,
US Iceman

Lc_shi
06-10-2005, 07:36 AM
Hi Iceman,
Your information is very valuable.
I've used Coolpack sometimes,it's rather good.But I don't know anything about EES. I think it's of great help if an engineer can compile some software for his own work. EES looks very good but expensive for me. I am not sure if Matlab can do something in RE field.

thx

regards
LC

US Iceman
06-10-2005, 04:49 PM
Hi LC,

I have not used Matlab, but have used Mathcad. The last version of Mathcad I have is about 5 years old.

At that time you had to write the programs to access fluid properties. or write the refrigerant equations into the actual Mathcad document.

Developing equation of state models to predict refrigerant properties for multiple refrigerants was a very involved task.

One of the programs I worked on was for large screw compressors. We eventually developed a software link ( a dll file) to use the properties from another program called RefPROP. (Available from NIST here in the US)

Refrigerant properties are required to fully understand the impact of dynamic changes in the system. You need them for pressure loss in piping, mass and volume flow calculations, quality, etc.

The other task is to develop the actual model of all of the system components (individually, and combined in the system).

I gave up using Mathcad except for some old files I have on shell & tube heat exchangers.

I have since switched to using EES as it is much easier and faster to use, since it has the refrigerant properties built-in.

All of the equations are written into the EES document using syntax similar to Excel. It is fairly easy to use and learn.

I have found some applications where spreadsheets work OK. Others are more involved and require the computing power found in EES. (at least from my perspective)

Excel has a built-in solver, but you still need the refrigerant properties. So, you either have to write the equation of state routines, use look-up tables, curve fit data, or develop software to link to the NIST dll.

All of these are not much fun!:(

Some thoughts for you to consider based on my past experience.

Regards,
US Iceman

Lc_shi
08-10-2005, 01:05 AM
Hi Iceman,
I totally agree with you. Refrigeration engineer is not a software engineer. I've once used FORTRAN to compressor simulation, it's really a tough work -:p. Usually we use the softwares worked by expert. Refrigeration engineer need pay more attention to the product development and final performance. Even the best software can't cover all the practical result of the design.
I'm grateful to you for sharing with me your experience.

rgds
LC-:)

US Iceman
08-10-2005, 01:39 AM
Hi LC,

It seems we are becoming more dependent on software to do our jobs. 50 years ago software was called a slide rule!

When you look at old pieces of equipment that ran for many years, you have to appreciate how the old engineers got so much accomplished.

We have many different kinds of engineers within refrigeration; product development, design, application, manufacturing, etc. Each one of these can become a field of specialization.

The one area I find the most interesting is application engineering. Using commercially available components to fulfill a specified requirement. One of the problems with this is the need for modeling of system performance.

So, we have to become software engineers to solve our refrigeration problems.


Usually we use the software worked by expert.

Software is only as good as the person who wrote it. And, if the error checking/trapping in the software is faulty, we get numbers in and numbers out. They can be the wrong numbers!


Even the best software can't cover all the practical result of the design.

I completely agree with you.

The software can do math and follow logic functions, but it is not smart. The person using the software still needs to understand the relative answers and the magnitude.

This is called experience.:D

This is a really good discussion.

Regards,
US Iceman

Lc_shi
08-10-2005, 04:21 AM
Hi Iceman,
Very glad to have discussion on this topic.
I've used slide rule in my first year in university and calculator come out the second year. I think who the age below 40 don't know about it.
Computer is main tool in our daily work. I can't imagine how an engineer does his work without computer's aid.To some extent,every enigineer is a software engineer,or software application engineer-:)at least.
I know refrigeration engineers here are categorized into:
# product development
# HVAC/R system design (equal to your application engineering)
# lab and test engineering
# installation and maitenance
# sales engineer
I think each requires some special talent. My interesting is product design.It's very challenging but many unfamiliar things made me headache-:p

glad to talk to you.

regards
LC

US Iceman
08-10-2005, 05:07 PM
HI LC,


It's very challenging but many unfamiliar things made me headache

I have experienced this also.;)

I do not remember all of the late nights I used to re-write software code, verify error checks, or logic.

There were times when the modeling consumed as much time as the software code. Then you have logic, syntax, program compiling problems, etc.

To solve the unknown can be fun and frustrating at the same time. Headaches and grey hair come with the job.:D

rbartlett
08-10-2005, 06:28 PM
I'm sitting here trying to get a little further with my windows Ductulator program for the sales reps to use with basic duct design. It'll be finished within a few weeks. I think I only need another 10 hours on it but 10 free hours might only come in scattered from over the next 2 to 3 weeks.

Then I want to work from spread sheet system modeling (SVT) to basic windows app System Variables Tracking.

My first SVT application will focus on the condenser and head pressure control strategies (philosophies). I find that many of my trouble shooting consultancy jobs ultimately come about from problems in this area.

I have been thinking of setting up another discussion board on Fridgetech.Com for exactly this purpose. Somewhere a handful of us can discuss the shared development of our tools without giving aware too many hard established secrets.

those old fridgetech boards were great. i recommend them to people and if they hunt the fridgetech site they can still find the archives

you should consider doing some more of the puzzles

cheers

richard

Lc_shi
09-10-2005, 06:09 AM
Hi Mark,
Pls give notice while you finish the discussion board setup. Where can I reach your first SVT? I want to have a try-:)

rgds
LC

Vincent Yu
10-10-2005, 05:09 PM
Hello LC and US Iceman,

I am interested in the topic you are talking about. I have the same problem as LC. My boss and QA department always ask me to give a relationship to predict the performances off the rating condition. For example, usually we give our products’ capacity at 68 degree F ambient temperature. But it is very difficult for us the keep the ambient temperature to be exactly 68 degree when we test them in production lines. We need a way to tell how much cooling capacities at different ambient temperatures, such as 69, 70, 71 … degree F, are equivalent to the capacity we give at 68 degree F. I do think it is too tough for me to give an accurate relation. Finally, I ask them to run some tests to get the relations experimentally.

If we could simulate the refrigeration systems accurately, our lives would be much easier. I have ever taken some time to think about it. I feel it is too complicated for me, and finally give up it. Maybe I approach the problem in a wrong way. By my understanding, there are two things we need to figure out. First, model the system in the right way. Second, use a solver to solve the mathematical model. EES just provides a tool to solve mathematical models. The most important thing is to set up the model, which should be simple enough to be solved by the tools we use and also accurate enough to the real problem. I have used CoolPack sometimes. The model is quite simple. It just uses mass conservation equation and energy conservation equation and holds a data bank of thermophysical properties.

Could you tell me which kind of simulations you can make? Can you simulate the whole refrigeration system or just simulate part of a system, such as simulating the work of a compressor or predicting the performance of a refrigerant mixture? If you want to predict the performance off the standard conditions, you have to simulate the whole system, how do you handle the heat transfer and the pressure drops in the condenser, evaporator, and the pipe? Do you use CFD or EFA methods point by point or just use the traditional way to give an overall heat transfer coefficient?

Please give me some idea about how to model a system.

Thanks
Vincent

US Iceman
10-10-2005, 06:48 PM
Hi Vincent,

A model is a mathematical solution of the interaction of the installed components.

In one case you can use manufacturers data for each component. You will need a range of operation and possibly several different conditions.

Let's say you curve fit the data points for a compressor into an equation.

The data should be acceptable for the range of conditions in question. For example, how do you determine the compressor inlet condition? If you use pressure, you have to include the effects of suction superheat (useful or non-useful).

You also need condensing temperature and the amount of subcooling.

Curve fitting all of this data into several equations can be difficult. Then you can have interpolation errors.

Or,

You can extract the volumetric and isentropic efficiency out of the manufacturers data and use this based on the pressure ratio of the operating condition.

Or,

You can develop the engineering principles of the component and use this (this is the most time consuming method).

Piping is the easiest thing to do. Use the Darcy-Weisbach correlation. However, when you have to solve for the friction factor you need the solver. This is where EES comes in. It just crunches numbers and solves equations. It makes short work of the friction factor calculation.


First, model the system in the right way. Second, use a solver to solve the mathematical model.

The first part is the hardest thing to do. You have to decide what you want to model, and, what the inputs are for the varying conditions.

Second, you need to decide how you want to model the components; manufacturers data or calculated data (possibly extracted data)

Third, you need the refrigerant properties in the program.

Fourth, you need to define the interaction of the components with the system (including the piping and the external effects of ambient weather)

Fifth and possibly just as difficult, perform the validation of the model. The validation and accuracy will be equivalent to the methods you choose. If you use a simple model, you should expect a greater deviation in the results.

This is not as simple as it sounds and I'm sure you will agree.

If you are testing the products at various conditions you may need an environmental chamber to develop the actual/desired ambient conditions.

Some ideas for you to consider.

Regards,
US Iceman

cool-hvacer
14-10-2008, 01:46 AM
This is a very cool thread......;)

It would be interesting to hear if there have been any new developmnets in this area in the last few years! I have been searching for software/simulation tools for HVAC/R for a while now without much luck (unless you are willing to drop big $).

I am not sure if it is really possible to make an entire model in excel. If you could, then almost everyone out there would be able to use it (doesn't everyone have excel??)

I do have access to simulink, but can't find any free libraries. I would have thought there would be a ton of them kicking around! I will probably just "suck it up" and make my own model....

If I do get one running, I'll post it!

Xooyoo
16-10-2008, 08:20 PM
This is my forst post here so:Hello!
I've been working on this issue for some months.

I know that that the biggest german company in the refrigeration area has been working for years! on simulating their products in dependence on time. But this is an extremly difficult and very expensive project. At least the program was calculating about 45 min on an actual dual core PC and still had very high tolerances. So far for calculating dynamic. The program was developed in Dymola Modelica.

I worked with EES quite a lot, it took me several month to describe an dx air cooled evaporator in static points. It calculated but had too less options and too great tolerances. So I got frustraded on old literature and wired mathematic dependencies, so I acutally do not move this project forward.
A main problem in EES is, that you're not abele to use object-oriented programming, and when your source code get's longer than 6 screen's you'll scroll to death. To do an really dynamic simulation in EES is also not very wise because it isn't built for.

I decided to learn OOP before I'll probably go back to the idea of simulate refrigeration components.

nh3simman
18-10-2008, 12:10 PM
I dislike these packaged solvers like EES and its derivatives like Coolpac. Not because they are bad but because they cramp your style.

If you really want to model thermal systems, then get a compiler and learn some numerical methods.

I like Delphi and I'm a big fan of the Numerical Recipes (NR) from Cambridge Uni. For the engineer who wants to program a solution and not spend years learning numerical methods it is fantastic.

But here is another option that I found about 6 months ago.

Python is a free programming language that I downloaded, learned and wrote a refrigeration simulation in 1 week. One of the authors of NR has even posted a big part of the NR library in Python.

cool-hvacer
26-11-2008, 02:44 AM
After giving this a bit of thought...... Are we making this a lot more difficult than we need to?

In reality, I do not need solutions to the "dynamic" problem, as much as I am looking for the new "steady state" solution.

For example, I have a chiller that is chugging away, and I change one of the loads or open a valve. I really don't care what the transient analysis looks like - I just want the new steady state.

I guess the "Rolls Royce" model would provide all of the details of the transients in real time, but who cares whats happemning after 1s, 2s, 5s, 10s, as long as I know that in 5 or 10 minutes, things have settled to my new solution.

Just a thought.......:confused:

autt
27-11-2008, 04:23 PM
In my mind 'dynamic' equals to 'transient'. A state that deviate/off the standard condition can still run steadily, a static calculation can solve this situation.

I had finished the static system simulation last year, and I think the dynamic simulation is not so urgent for general using.

I'll post a new thread to help test my model.

jinbei
11-04-2010, 04:39 PM
hi all..
i'm new here..
i have a task about mathematical modelling with the topic: optimization refrigeration system design.
of course with safety and cost as the concerns.
i wanna ask,
1. how do i start this modelling?
2. what is the "objection function" i can use to work with this model?
3. and the goal is:
based on the model, how should i put the
a. compressor
b. condenser
c. Receiver
d. Heat Exchanger
e. Suction drum
in a place with high safety and minimum cost?
with space between each instruments and safety number as the results.

i'm not an engineer, i'm a mathematician so i ask from the basic.

thanks before.. :)