Flow through an actuated Valve

Hi All,

Does anyone have a lead for a formula to calculate flow through an actuated valve? I recall discussing this before, but not finding anything.

Some will scream at my pressures and hip shooting due to some recent thread discussions, but this is what I have and am targeting.

Valve duration, seat to seat is approx. 50 deg of rotation. Calling 30 deg effective @ 900 rpm/15 rps/.066 sec cycle duration, I come up with ((30/360)*.066)=.0055s effective duration. Therefore I get 1/.0055=181X flow requirement. My desired flow volume is appox. half of .88in^3 or .44 in^3/cycle. .44*181=79.64 in^3/s. Is it possible I'm in the ballpark if I vary the pressure a notch?

The units in the calculator are slighly elusive, but i changed variables to hit known hard targets. Some remain elusive. Thanks for any review effort, Keith.

Hey Keith,

These two papers are excellent.

Steam flow through the inlet and exhaust arn't static, they are dynamic, so if variables are left out then the calculations will be in error.

Caleb Ramsby

Thanks Caleb, nice data and formulas. (I prefer your adjective "excellent", worth an edit.)



Edited 1 time(s). Last edit at 03/02/2015 01:47PM by kdc2.

Modeling valve flow is very complicated. I have been looking for simpler ways of doing it. A partical model would be the most accurate. But vary compute intensive. I have thought maybe, i'll call it a bubble model, might work. Something like a partical but larger and having variable volume.

Andy



Edited 1 time(s). Last edit at 03/07/2015 02:22PM by steamerandy.

Andy, Indeed a complicated flow model. The larger picture has both more complications, but nice conveniences too. I must limit this expander to 1000 psi. So I will back into TDC and any lead on the fly if measurements allow. The steam generator is supercritical, necessitated by design, materials on hand and desire to prove a compact package. There really is little temperature losses in throttling and huge gains in superheat, so I have a nice cushion with an upstream target of 2000 psi. It is compounded to ensure there is no quick and easy accurate flow model.

Keith

A few charts, very far removed from perfection

Caleb Ramsby

A few more charts.

If I recall correctly the first charts, A through C, are just over 10% cutoff. D is for just over 2% cutoff.

These only show the head end, I also do the crank end of the piston, the torque diagram showing the effect of both ends, double acting simple.

The charts all show the production of roughly 20 hp for a motorcycle speed of 80 mph. Some with the relief check valve, (ala. Stumpf or Williams) some show the independent exhaust valve.

I somewhat expanded upon Professor Hills formulas, except I havn't added the thermal heat loss to the cylinder. Gotta get the flow and expansion/compression principles correct for the model first.

One thing I can add, is that using a cross check for the expansion of the steam after valve closure of the weight of the steam in the cylinder divulges some curious goings on in regards to the coefficient of expansion.

Caleb Ramsby

Ah, dug up my notes.

Inlet superheat at 800 F.

Chart:
A, 2,799 rpm, 20.79 hp, 11.39 lbs hp hr, 477 psia inlet.
B, 2,799 rpm, 20.66 hp, 12.56 lbs hp hr, 300 psia
C, 1,529 rpm, 20.68 hp, 12.05 lbs hp hr, 447 psia
D, 2,799 rpm, 20.77 hp, 7.63 lbs hp hr, 1,300 psia

This is for a 2 1/4" bore by 2 3/8" stroke three cylinder double acting simple, semi-unaflow. With 1 1/8" inlet piston valves, 5/8" piston exhaust valves and a 3/8" diameter poppet style compression relief valve.

I have the various engine frictions being accounted for and on top of that use a card factor of 85%. I need to implement leakage now and refine the friction analysis for the variance in coefficient for the piston and valve rings through the stroke. Lots of work left to do, so take the results with a SALT LICK!

Caleb Ramsby

Thanks for the efforts Caleb.

My crude model, just displacement based, has the a side at 16% cutoff but the b side is 26% so that when combined with compounding and compression I hit my target of 1000 psi. Rod is a pretty significant volume displacer in this mini proto. I can see implementing the bi directional upstream/downsteam calculations will be a mofo. It will need to be done. Trying to prioritize the model, the build and running testing. Can't beat running hardware to verify models. Praying the 500+ deg of superheat will get me through condensation issues. Keith

Hey Keith,

Yeah, this isn't easy. It has taken a lot of time and effort for me to be able to produce these graphs, but it is still not anywhere close to where I want it to be.

Modeling the connecting rod is mandatory for analyzing the valve flow with any significant degree of accuracy. Likewise modeling the valve gear, whatever type it may be, is mandatory for accuracy. For the above diagrams the inlet valve is driven by the Walschearts type and the exhaust valve by the Joy type. . . don't ask why! I need to put them both back to using the Joy gear, the 5/8" lap and 1/64" lead I am using just isn't practical with the Walschearts gear when the engines stroke is only 2 3/8", very bad valve gear geometry produced from those proportions. The Joy is more versatile and less mechanically involved. The Stephenson is too bulky to drive both the inlet and exhaust valves independently. The cam type just isn't practical for an inline three with the variation of cutoffs I want to run without going to Caprotti type or something similar, too involved and of dubious advantage overall, in my opinion at least.

The spreadsheet uses 10 steps of calculations per degree of engine rotation for all 360 degrees, doing the head and crank end of the engine simultaneously in separate columns. I use a manual input feed back loop for the initial pressure and temperature of the intake stroke of the head end and start of the exhaust/compression stroke of the crank end of the piston. Simply alter the manual input until it matches the end calculation for that piston end. Since the last calculation for the head end is the final compression pressure and temperature of the steam at TDC, for the crank end the final calculation is for the opposite condition. That is the pressure and temperature when it is at the end of its inlet/expansion stroke. The spreadsheet does the rest.

I still havn't gotten around to adding the effect of the steam line coming from the boiler and the steam chest in the variance of the effective steam chest pressure stability. . . still a lot left to do.

Like you said, testing real metal engines with live steam is the only way to get truly accurate results. Determining what design to build and test. . . well that is what I have been writing this engine analysis spreadsheet for.

Have fun, try not to lose your mind!

Have you investigated Professor Halls program that he wrote based on the papers he wrote?

Totally agree with the need for modeling Caleb. I spent months working with the timing of valve events. I was able to keep common piston rods and shafts, but I even stroked the LP cylinder a bit longer for valve timing events. Valve gear is cam driven hydraulics, too complicated with the inertia problems for me to implement a typical valve gear. Have a really cool reversing setup but am totally focused on stationary apps at the moment. Much easier total system. Mobile has to wait. Keith

Keith

I suffer from both lexydisia and dyscalculia. These are significant burdens, given what I do for a living. Be warned- don't trust me with numbers. However, I do have a pretty good head for concepts.

The orifice2000 program you are using was designed for measuring flow through pipelines using the pressure drop across an orifice. The orifice is round, has a 90 degree knife edge on the upstream side and is very thin compared to the diameter of the pipe. The pipe is assumed to be mill finish. Many diameters of straight pipe are assumed upstream and down stream to "straighten the flow." And of course, pulsation or intermittent flow throws the whole thing out of wack.

The expressions "orifice taps" "corner taps" or "pipe taps" refer to the exact locations where differential pressure is measured. Orifice or corner taps are located very close to the orifice while pipe taps are further up and down stream. As it turns out, the pressure rises immediately upstream of the orifice, drops sharply going through the orifice as velocity increases and then rises again further downstream as velocity decreases. Orifice taps or corner taps give a higher differential pressure other things being equal. Pipe taps measure the permanent pressure loss due to friction through the orifice.

The basic orifice constants were determined empirically with a surprisingly small number of experimental set ups, and usually with water. Ideal gas and liquid laws were assumed. Then equations were sought out that fit observation, and a number of correction factors were applied. This all works very well if one is measuring flow through pipelines.

Formulae are available which predict the flow through a valve. The catch is, you have to know the Cv of the valve. This is determined empirically by whoever built the valve. A valve with a Cv of 1 flows one gallon per minute water with one PSI drop. Similar Cg and Cs values apply to the flow of gas or steam. Valve calculations are a starting point, especially when you are using steam or gas. A wise engineer expects that the valve might not flow as much as calculations would indicate and incorporates a fudge factor.

Bottom line:

The program you are using is useful as a guide, but do not expect accuracy. You are extrapolating beyond applicable conditions. You are just going to have to build one and see what it will flow.

Kerry

Thanks Kerry and I agree whole heartedly. Thankfully I can take that 2000 psi up to 3500 to try and target my flowrate. I am confident it should run well enough to collect data and replace parts as needed. Most all components have about a 5x safety factor built in. I'm trying to wring out as many variables as I can/is practical prior to operational testing. Keith

I use VisSim in most of my analysis now. I have talked about it befor. It is a general mathematical simulation package that simulates time dependent processes. Time is the independent variable.

A full engine simulation would require a load simulation on the output shaft. Basically a dyno simulator. The engine simulator would then need to calculate the shaft torque. Cylinder pressure to force on piston transmitted to the output shaft torque.

My progress was slowed when I upgraded to a new version of VisSim. Had to rewrite a lot of the VisSim plug-in function interfaces. Most is working great. But my hard drive failed so saving for a computer replacement at the moment. It was a bit outdated anyway. The Note 4 I am posting with right now has around 4 or 5 times the compute power of my old computer. To bad it's not compatabile with IA86 Windows.

VisSim is actually a graphical programming language. A block is a function having inputs and outputs. Graphicly optputs are conceded to inputs. This is exactly the way old analog computers were programmed. Except the blocks are now graphically represented. It is easy to use if you have mathematical background. It has many blocks to emulate mathematical functions. A sum block for example outputs the sum of its inputs. An integrator block simulates a mathematical integral function. A mechanical equivalent is a spring. Or electrical a capacitor. A model can be a part. You can select blocks and create a sub block out of them. Just in other programming languages you would write sub-programs or functions that can then be used in an number of places. But here we have a different perspective. A block might be an expander simulating the steam flow into and out of the cylinder and steam expansion etc. These may also may Also may use user created blocks. A valve simulator block for example. All the things may need have constant parameters. Creating a blocks allows parameters to be specified that can be set for each 8 stance of a user created block. Blocks may also be implemented in a programming language. These blocks become built in blocks you can place in your simulation. The steam property blocks I have programed are written in C++. I have extend the IAPWS properties to include derivities of properties along constant entropy and enthalpy paths allowing constant entropy or enthalpy processes to be simulated. The IAPWS 95 Scientific Formulations all have temperature and density is the input independent variables. The rate of change of density with respect to time Dd|s entropy s constant in the cylinder can be calculated from the piston position, velocity and other mostly constant cylander parameters. Multiply dT/dd|s × dd/dt you have dT/dt|s. Running that thru an integration block and you get T, temperature, to goto into a steam state point block. Integrate the dd/dT to get density and you have a constant entropy change. Develop a model for heat transfer in the cylinder that gives a rate of change of temperature that can be sued withe the constant entropy change to simulate the cylinder steam state during expansion and compression. Every component can simulated and made into block. Then an engineer can be simulated by putting the blocks together. The blocks will need to be tested against an actual working engine. That is my interest in pressure transducers. My thinking is that if a simulation is correct the cylinder pressure should track actually engines. Other things like steam consimption would also have toatch. Comparing simulation generated data with actual instrumented engine data is the only valid proff. And even then one cannot be sure it is general enough for different engine designs. Lots of testing needs to be done.

Andy

Andy,you done good ,my old windows 98 and similar shop and steam piston routines won't even load on windows 7.
If you can get to the point of making them run again...the entire steam community would benefit from it.

There is a lot of information on poppet valves for internal combustion engines. The intake valve(s) in a four stroke engine are critical. A standard work is "The Internal Combustion Engine---" by C. FayetteTaylor et al. Chapter 12 deals with valves including equations for flow. What's interesting to me is the lack of similar study of flow in reciprocating steam expanders. The pulsating nature of flow, especially in two stroke IC engines, resulted in an intensive study that more than doubled the power from an engine of a given size. Flow control in two strokes directly determines scavenging efficiency. This is true also in four stroke engines where tuned intakes and exhausts have been used for years, Four stroke valves are aimed to give controlled amounts of tumble and swirl to help with lean burn fuel economy as well as power.

A very long time ago it was realized that a flow index related to the speed of sound in intake valves was related to volumetric efficiency in IC engines. That limited the piston speed for a given valve area. This should be true for steam engines. Also the lift in IC engines is usually around 0.3 times the diameter. Steam poppet valves seem to have a lot less lift. Some of this is probably the very short open duration which requires a large cam base diameter for even small lifts.

I see a lot of thermodynamic analysis here without much emphasis on the details of steam flow in real engines. IC engine simulators have been available a long time. The late Gordon Blair developed some of the early computer based simulations and published his work in book form. Some first degree simulators based on his work are available at low cost. They deal with combustion simulation as well as flow and expansion. A reciprocating steam expander simulation would only need to deal with flow and expansion, both fairly well understood compared to combustion in an IC engine. It sounds like some of these simulations have been done for steam expanders.

I found three NACA reports/notes on the subject of air flow through a poppet valve. I would be interested in why superheated steam flow wouldn't be similar.

NACA report #24
NACA tech note #915
NACA tech note #1035
available at [naca.larc.nasa.gov]

Lohring Miller

Hi Lohring,

Any references to document ID, publish date, Report/Patent Number? Tech notes and report # don't seem to be searchable catagories?

Looks like there is a lot of info available.

Keith

I have them pretty much working in single state reagions. I havn't figured how to figure out the rate of change properties for a mixed state. I was working on a way of figuring that when my hard drake quit. I have most of my file backed on an external drive.

It's amazing how fast 1000s of points can be calculated when you have a way to calculate the dependent properties from the previous state. The numerical integration isn't the fast of calculations. But it is a lot faster then solving for two dependent property for a given set of dependent ones. At least with the IAPWS 95 formulations you have density and temperature for the independent properties. That is all other properties are calculated from given temperature and density. And sense density I can be calculated from cylinder volume and steam mass. One need only get the temperature. The IAPWS formulations are available on the IAPWS Web site.

ANDY

The progression from the early report through the tech notes is an interesting walk through the history of engine development and testing. I'm still debating if air flow at the comparatively low temperatures and pressures would be the same with superheated steam. It should act like an incompressible gas with the (hopefully) low pressure drop through a valve, shouldn;t it? Unfortunately the .pdfs are too large to attach.

Report 24 is titled Air flow through poppet valves
NACA tech note #915 is titled The effect of inlet-valve design, size, and lift on the air capacity and output of a four-stroke engine
NACA tech note #1035 is titled Steady- and intermittent-flow coefficients of poppet intake valves

A search on the titles should work.

Lohring Miller



Edited 2 time(s). Last edit at 03/10/2015 08:38AM by lohring.

All files from this thread

File Name	File Size		Posted by	Date
2000 orifice.png	67.8 KB	open | download	kdc2	03/02/2015	Read message
Predicting Locomotive Performance.pdf	148.2 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Predicting Performance - the Problem of Condensation.pdf	37.1 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator A1 card.JPG	101.3 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator A2 torque.JPG	134.1 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator B1 card.JPG	100.3 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator B2 torque.JPG	126.1 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator C1 card.JPG	101.7 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator C2 torque.JPG	120.8 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator D1 card.JPG	106.7 KB	open | download	Caleb Ramsby	03/02/2015	Read message
Indicator D2 torque.JPG	126 KB	open | download	Caleb Ramsby	03/02/2015	Read message