Control.com - Nerds in Control

Hello, similar thread about 5 and 7 element drum control made me thinking about how to develop boiler drum model that will represent drum dynamics around some operational point.

For example, if you need to perform system identification, first you need to choose input and output variables. I have seen various drum models, but I wonder if it is possible to model only drum level. For example, to try to amke model with the following input variables:
Feedwater flow, Steam flow out of boiler, Pressure of feedwater and output varible is water level in a drum.

Since 3 element control that is often used measure three variable: drum level, feedwater flow, and output steam flow, I think it is sufficient enough to use these variables + feedwater pressure to obtain fairly accurate model that will good describe dynamics arounf operational point.

What do you think? Will this be enough?

Hello Mik,

I am just looking at a similar problem to yours: modeling a (shell) boiler, including shrink/swell effects. Unless you are developing a very simple model, you will need as inputs:
Feedwater flow
Steam flow out
Feedwater enthalpy (or temperature)
Heating energy flow (or fuel gas flow or other fuel flow)

I do not think you will need the feedwater pressure; however, you will probably need to use/calculate the (steam) pressure inside the boiler.

Two articles on boiler modelling that you could read are:

S. Lu, "Dynamic modelling and simulation of power plant systems", Proc Instn Mech Engrs, Vol 213, Part A, 1999, pp. 7-22.

K. J. Astrom and R. D. Bell, "Drum-boiler dynamics", Automatica 36 (2000), pp. 363-378.

If anyone has better or more suggestions for modelling boilers, I would like to hear about it as well.

Best regards,
Gerrit

Thank you for your reply. I know that pressure inside a drum is correlated with output steam flow. On the other hand, I don't know why to use fuel as input if I'm not going to model how pressure is changed. I think that steam flow will be sufficient to see how fuel (indirectly) affect level.

BTW, do you know where to obtain this article from K. J. Astrom and R. D. Bell, "Drum-boiler dynamics"?

Regards

If you are doing a very simple model of the level, you can assume that the steam mass in the boiler and the water density are both constant, that there is no vapour under the water level, etc. Your model then becomes:

dV/dt = (wfeedwater - wsteam)/rho
L = f(V)

where

V = liquid water volume
wfeedwater = feedwater mass flow
wsteam = steam mass flow
rho = water density
L = water level
f(.) = function relating water volume and level; this depends on the
boiler geometry, e.g. a vertical or horizontal cylinder

However, the above assumptions are very simplifying. For a more realistic model you will need to take into account that the steam and water densities vary with pressure, which depends on the fuel flow and the feedwater enthalpy (temperature) as well as the feedwater and steam flows, etc...

The article you mention, with a fairly complex model, can be found on:

http://www.ingentaconnect.com/content/els/00051098/2000/000 00036/00000003/art00171

If you are not with a university, you will probably have to pay to access it. If you have a university with a reasonable engineering department nearby, they may subscribe to Automatica, in either paper or electronic form.

Best regards,
Gerrit
--
*****************************************************************< br>Gerrit M. van der Molen, MSc, PhD
ISC Ltd.
************************************************************** ***
50 George Street, Glasgow, G1 1QE
United Kingdom
*********************************************************** ******
Tel: +44 (0) 141 553 1111
Fax: +44 (0) 141 553 1232
Email: gerrit@isc-ltd.com
Web: www.isc-ltd.com
********* ********************************************************

Hi,

Which software are you using for simulating the drum level control?

Energy input is very significant, as mentioned.
Swell and shrink effects are very significant, especially in a poorly designed boiler.

With the increase in energy, specific volume of the water inventory changes, thereby very quickly changing drum level, without actually changing the water inventory. (more bubbles in the evaporative circuits).

I am also interested in system identification.

Kind Regards,

Gerrie Visser
Babcock SA

Well,
I'd be able to obtain data from one coal power plant with steam drum. I intend to use system identification toolbox software. I need to specify which variables I need in order to obtain satisfactory model, at least near working point in steady state. I thought to use only one output, and that is drum level.

Measurement is done by differential pressure transmitters. Since drum pressure and steam output flow is directly correlated with amount of fuel, I thought to use these instead of input energy, because that would be hard to calculate.

Well if you measure fuel input on a PF boiler, you can get a good estimate of heat input - so no that will do fine. If PF, swell and shrink effects should not be that great, as you would expect a utility boiler to be a baseload unit. So with system identification, you could get good results, linearised about an operating point.

I am using Scilab, Scicos - similar to Simulink.
It is free, but I am not sure how good the system identification will be.

Gerrie