Steam drum boiler model

Micik wrote:

Steam quality relates to liquid water carried away entrained
in the steam as droplets.
http://www.plantservices.com/articles/2003/378.html

Low quality steam implies that more water mass is leaving
the drum than you would expect just from the mass of steam
being produced.


I don't think so.  I think riser wall refers to the tube
wall that is the main heat exchange surface of the boiler,
where water boils as it rises from the lowest point in the
water circulation system (usually a lower distribution
header below the drum) and reenters the drum near the drum
water surface level.  The risers contain a mixture of water
and steam, so their surface temperature can be above the
drum temperature.
See:
http://www.spiraxsarco.com/resources/steam-engineering-tutorials/the-boiler-house/water-tube-boilers.asp


I haven't seen this model, and it will take me quite a while
to try to digest everything in this article.

It worries me that the model does not appear to contain (if
I am understanding what I an seeing) a term for the make-up
water temperature (or the difference between the drum
temperature and the make-up temperature).  This factor has a
strong effect on boiler stability.

--
Regards,

John Popelish

I looked quickly.  Usually quality is used to express the proportion
of saturated steam.


This is the first I have seen for a boiler.


I would use a model like that if I were designing boilers.  I think
mathematica should have picked a simpler example because no one is
going to work their way through this example unless they have a need
to design boilers.  The example is poor because the different states
are clearly explained.  There must be more information for this
example somewhere that we don't have access to.   A simple example
were a hot and cold water valve or pump controls the level and
temperature in a tank would have been better.

I do modeling all the time.  I was working on a modeling problem this
weekend.  I usually start simple and add to it as the need for better
simulations arise and time permits.  I have done hydraulic simulations
where there are equations for the pump, accumulator, valve spool
position,  flow,  the change in pressure in either side of the piston,
force, acceleration, velocity and finally position.  I can't use state
space because the system is non-linear.    All this is necessary and I
still don't account for the pressure drops due to length of piping.

It is safer to use models to prove a system will not work.  I have
saved customers a lot of money by just doing this.  I don't think it
is safe to use a model to predict a model will work unless you are
very sure of the model.  If I do give the design the OK I usually can
see exactly what is going to be the factor that restricts
performance.  Too many hydraulic designs fail because components were
just thrown together and they expect the motion controller to correct
for design deficiencies.

Peter Nachtwey

On Mon, 28 Apr 2008 07:46:17 -0700, pnachtwey wrote:

-- snip --

I would completely agree with that last sentence if you would just take
out the word "hydraulic".

It's amazing how much money you can spend on a design project by not
doing it right the first time around.

--
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html

u6zaYjVnZ2dnUVZ_jWdnZ2d@web-ster.com:


Have you ever looked at the FDA waterfall approach to medical device
design?  These days, its almost mandatory if one wishes to seek FDA
approval.  It's not such a bad exercise.


--
Scott
Reverse name to reply

On Tue, 29 Apr 2008 00:48:39 +0000, Scott Seidman wrote:


I have not.  If it's a pure waterfall model then it'll work well if
you're designing something that you already know how to build.  This
would be consistent with what I've been told about life-critical design
for medical and flight software -- you get it working as an experimental
prototype, archive your design onto difficult-to-access media, and do it
again from scratch the 'right' way.

Most design efforts that I've been involved in are either "stumble around
spending money and time while management plays head games", or are the
'spiral' model, where you do ever-descending 'mini-waterfalls' until you
finally have a product.  The former is generally good for whoever gets
promoted before the fertilizer hits the fan*, the latter is good for
getting a pretty darn good (but not entirely bug free) product to market
in a reasonable amount of time.

* Did I mention that I'm no longer temperamentally suited to being an
employee?  It's a good thing I have skills as a consultant...

--
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html

Worth a look

http://www.fda.gov/cdrh/comp/designgd.html





--
Scott
Reverse name to reply

Micik wrote:

I think that Peter's reply touched on a lot of the points that I would
make, but I want to put a different spin on it:

There is no one right way to model any given system, or part of a
system.  No mathematical model is ever 100% right, and the more accurate
you make your model the more expensive it is to design and maintain, and
the more accurate your model is (usually) the more difficult it is to
analyze it or simulate it.

So you should never try to make a model that is 'good' in a global
sense, you should only try to make a model that is accurate enough for
your purposes.  Get it too accurate and you're wasting time, don't get
it accurate enough and you're wasting money (and time, and aggravation,
and, and, and).

For example:

If you are doing preliminary design on your control system, you should
model things accurately enough so that you know what sort of valves,
actuators and whatnot you need to use; at this stage of the game your
model only needs to be accurate enough to give you a fairly conservative
estimate for the speed and accuracy of your actuators and sensors, and
the sampling rate and complexity of your control rule (which will
determine how fancy of a controller you need to specify).

If you're going to have a chance to work with the physical plant before
the whole system is commissioned (or before your product goes into
production, if you're building a production machine), and if the health
of your plant and the people in it's vicinity doesn't depend on the
controller, then you may be able to abandon your modeling effort at this
point.

On the other hand, if your controller has to be right the very first
time (such as with modern fighter jets, which can't be flown unaided by
human beings), then your model may cost more to produce than one item of
your plant.

Note, too, that your purposes may be completely different from the guy
sitting next to you.  You may have a model of a steam boiler that took
you a man-year to develop, that will let you simulate not only normal,
but all sorts of abnormal operating modes without leaving your desk, and
that will let you do all the control system design in the world.  Yet,
if someone comes to you and asks how think the concrete needs to be
under the supports for the boiler, your model may not be able to provide
the first clue about how much the darn thing will weigh.  So _you_ may
have the Most Elaborate Model In The World, while the engineer next to
you may have one, too, of the same thing, and the models may be
_completely different_.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

If x3 and x1 vary independently, there are only two possible reasons - (1)
no boiling, and thus no steam, and (2) superheated drum contents, which
implies non-equilibrium behaviour and *very* complex behaviour.

Steam quality as I know it is the percentage of vapour in the steam. I'm not
sure how a thermodynamic model will derive this, it's a characteristic of
the separation internals within the drum. Empirically, it drops off if the
steam draw is very high.

The analysis focusses on sine wave response and stability. The real issue is
not that, practical boilers can almost always be controlled tolerably while
the load is high. The problems occur when load drops off, the poles shift
closer to the imaginary axis and response gets more and more sluggish and
oscillatory.