Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Flow Restrictor Model Equation

Status
Not open for further replies.

stroupaloop

Mechanical
Aug 1, 2008
23
0
0
US
I have been thumbing through the forum posts lately and figure that now is a as good of time as any to begin discussing different modeling equations necessary to calculate various fluid conditions of a system. I suppose coming from a mathematical background and my shear lack of faith in much of anything without analytical calculations justified why I enjoy mathematical modeling of the system. All calculations done will be at a steady state condition due to the introduction to non-linear differential equations required for dynamic analysis.

These forums are open to discussion and I look forward to people's feedback. I will attempt to post a new modeling equation when I have time.

Flow Restrictor Model Equation:

As we should all know as engineers, there are three types of flow conditions: laminar, transitional, and turbulent. Since the transitional phase is so unpredictable, we will only focus on laminar and turbulent conditions for now.

General Flow Restrictor Model Equation:

Q = K(r)*A(r)*deltaP^n

where:
Q = flow
K(r) = flow coefficient
A(r) = flow area
deltaP = pressure differential
n = exponent (0.5 for turbulent and 1 for laminar)

As everyone can see, to complete the equation, the remaining variable becomes K(r) which varies depending on the state of fluid being either laminar or turbulent.

Turblent:

K(r) = C(d)*(2/rho)^0.5

where:
C(d) = flow coefficient (assumed to be 0.61 from experimental data)
rho = fluid density

Laminar:

K(r) = D(r)^2/(32*mu*L)

where:
D(r) = flow diameter
mu = dynamic viscosity
L = restriction length

It is important to note that this equation is a general model for flow restrictors. An assumption that a sharp edged orifice is used. As the orifice changes, so does the design parameters on the equation. Additionally, flow will never be specifically laminar or turblent which means your "n" value will typically be in between 0.5 and 1. Both "actual" changes in phsyical representation of the equation requires experimental data to verify such results, but this equation is best suited for modeling general steady state parameters.

 
Replies continue below

Recommended for you

What is r in the (r)?
What is d in (d)?
Are they function parameters, arrays indexes?
Do you mean?
K[sub]r[/sub]=C[sub]d[/sub]*(2/rho)[sup]0.5[/sup]
I haven't figured out make greek characters yet. I some one know can you share it with us? You should be able to look at the web page source to see how I made the superscripts and subscripts.

What is the name of the equation for laminar flow.
what does your equation do that this one doesn't:
 
Any mathematical model of a hydraulic component has so many assumptions that modeling a real physical component are close to impossible. All valves will have turbulent and laminar flow at the same time. If you are using a cartridge valve in a manifold it is even more difficult even with CFD to obtain a valid mathematical model. Choosing the correct compressibility of oil can have a significant impact on modeled results compared to actual results.

Ed Danzer
 
Any mathematical model of a hydraulic component has so many assumptions that modeling a real physical component are close to impossible.
Don't let perfection be the enemy of the good. Do you know that we can auto tune a hydraulic system? We use system identification to find a best fit model. The model isn't perfect but it is more than good enough. Auto tuning is done by comparing the response to the control signal. All this is done empirically. The problem with this is that the system is already built and it is screwed up, it will cost a lot to fix. The goal should be to model reasonably accurately and simulated before the system is built to prevent costly errors.

I was at a IFPE show many years ago. I attended a presentation where a Caterpillar engineer told us how they test components so they no more about them than the manufacturers do. The engineer showed the audience the model of the valve knowing darn well that it would blow everyone away. I am sure it did. Even I thought it was over kill. What I got out of the presentation is that modeling can be done.

The equations that Andrew has presented are simple. One should be able to emperically find the constants but it would be much better if the manufacturer would supply this information.
 
I apologize for my lack of presentation on the equation side, but the reformat of what you presented is correct.

The name for the constant equation used for the flow restrictor model is named:

K<sub>r</sub> = restrictor flow coefficient

which is dependent upon flow and fluid characteristics

As far as the relationship between the link you posted and the equation set here. This equation set is more focused on fixed orifixed, specifically sharped edged orifices not through tubes. I see that the equations are more focused on fluid velocity, but I will try to post something regarding what I use for determining pressure flow characteristics across a length of rigid tube.

Regarding the accuracy of modeling equations. I've done enough consulting jobs that I've see the power of consulting and the capability of what they can do. First off, it is very important to note that even the best system modeler with the best software can never replace test results. With that in mind, with enough experience, one can get very close to predicting the dynamic performance of a system by using the correct dynamic equations and iteration process. The company I used to work for, FES-BarDyne, Inc. uses a software named HyPneu, which simulates hydraulic and pneumatic circuits. Although the inteface sucked to high heaven, if you could get past that, which was very hard for me to do, the engineering computing capabilities were phenominal compared to Easy5 or Automation Studio. After a while, I would laugh when people would talk about results used from Automation Studio because I've used it and although its aesthetically pleasing, it still needs a lot of work on the analysis side.

Regardless, Caterpillar is a great example of how they can know exactly what they are getting and how its going to perform in their system through dynamic modeling so that when they produce 1,000 trucks in first line production, they know how its going to perform. Just takes time and money...

 
I forgot to add something regarding Peter's comment on auto-tuning. One of our current and most used fighter jets in today's warfare utilizes simulation software to determine leakage and system failure of their hydraulic system. It's named Model Reference Leak Detection Method, whereby a working dynamic model is created for the hydraulic system and through sensors that are carefully placed, the values are referenced back to an on-board computer that checks for a certain percentage of deviation. This deviation will flag any types of failures and will take corrective measures.
 
Functionally correct mathematical models of most cartridge valves will be like what the Cat guy described at the IFPE. I have found one Simulink model of an actual cartridge valve, a Sun counterbalance valve. The Sun engineer described it as very difficult to create this model.

I have purchased Matlab and Simulink, but need more tools (SimHydraulics and others) to continue trying to simulate my new motor concept. The manufacture of the high speed pilot valve cannot supply a mathematical model, nor can HydraForce for the larger control valve. I will be doing testing at different flows, pressures, temperatures, and back pressures to generate enough data points to make a good Simulink model as the response of the valves is one of the most important requirements to make the product function correctly.

Auto tuning a system is not a new concept, electric servo systems have done this for years and the principles will be similar. Simulating a design instead of auto tuning will only be done if you have no choice because of the difficulties in defining the components mathematically and setting up the model with all the components correctly is far more time consuming. The comment about capacitance in another thread is just one of the many hurdles in doing simulations correctly. The material used for the valve manifold will affect the capacitance of a system also. I always recommend carbon steel as it is stiffer than aluminum or cast iron (less capacitance).

Ed Danzer
 
Status
Not open for further replies.
Back
Top