Equipment, Piping, and Valve Mistakes Part 1 Tips

With a little help from my friends I have come up with a list of process, mechanical, and piping design mistakes that have made our job as automation engineers more challenging and in some cases impossible. We learn the most by our mistakes. Besides preventing the reoccurrence, this list should provide insight to prevent conceptually similar problems.

Solutia retiree and Fellow Hall of Famer Terry Tolliver and ISA Mentor Program participants Hector Torres and Hunter Vegas have helped me with this first list.

(1) Boiler drum that is too small for the steam production rate. Shrink and swell can be so severe for changes in steam demand that the boiler may trip on low or high level. The normal 3-element control that is enforcing the material balance by feedforward where the feedwater flow is ratioed to steam flow can make the initial level swing worse. A smarter feedforward uses a kicker algorithm to create a transient feedforward of the opposite sign as the normal 3-element drum feedforward. For an increase in firing rate for an increase steam demand there is an immediate temporary step decrease in feedwater flow to help mitigate the swell. The increase decays out as rapidly as the swell. The conventional feedforward can step in once the swell is under control.  The swell develops from an increase in bubbles in the downcomers from the increase in firing rate that pushes level up into the drum. Note that differential pressure level transmitters see the increases in head from the liquid but not the bubbles pushed up into the drum.

(2) Gas pressure control volume that is too small for gas flow rate. Rapid increases in gas flow can activate pressure relief devices. An extreme example was an incinerator where increases in waste gases could cause the pressure to ramp off-scale within a tenth of a second. In this case, the 0.1 second scan time of a DCS PID was too slow and we needed to revert back to the analog electronic controller. The fan had a variable speed drive with no dead time or dead band and the transmitter damping was minimized.

(3) Temperature or pH control in a pipeline with a long distance between the sensor and control valve. The transportation delay is excessive. Feedforward and dead time dominant tuning can help but the real solution is to move the sensor closer.

(4) The use of a control valve that is too big or a butterfly is used to control low flow. The valve ends up riding the seat. The equal percent characteristic is steepest and the sealing fraction is greatest near the closed position. Plus there is the breakaway torque when the valve closes. The result is a limit cycle near the closed position and on-off control that wears out the valve and increases system variability. This is a common problem for pH control because of the extreme rangeability requirement. The solution is a single correctly sized sliding stem valve with a positioner. To achieve greater rangeability without sacrificing sensitivity, large and small control valves can be installed in parallel and a combination of split range and valve position control (VPC) used. When demand is high, the splitter block output to increase large valve position is added as a feedforward signal to the VPC. Conventional flow feedforward and directional rate limits on the AO blocks with external reset feedback helps the VPC deal with load upsets.

(5) Large oversized blender discharge valves. They affect upstream and downstream feed systems. If a PID is controlling the blender's level it will react abruptly to every discharge and to the sudden large decrease in level. The originated sudden step changes to the upstream feeding system and subsequent decay in rate demand could affect raw material ratios and concentrations, especially if the associated feeders control loops are not properly tuned. Downstream the large amount of material being discharged will create stagnation zones and blocking when the discharge chute and vibrating conveyor are not large enough. Tuning of the level controller in the receiving hopper will be a hard task as stagnation and blocking will introduce constant load upsets varying in period and amplitude. Aggressive tuning leads to overfill of the hopper and high torque of the associated agitator; when the tuning goes to the other extreme, the material will travel slowly thru the conveyor leading to the mentioned blocking in the discharge chute or conveyor. While scheduled and adaptive tuning may help, the real solution if the process material permits is the replacement of the slide-gate with a properly sized rotary airlock valve (rotary airlock feeder) to provide a modulated continuous flow.

(6) Control valve upstream of flow sensor. The control valve distorts the flow profile and can cause bubble formation from the flashing in the vena contractor of the valve. A Coriolis meter can deal with the irregular profile and some degree of bubbles in special models but the real solution is to have the flow sensor upstream of the control valve.

(7) Steam desuperheater temperature sensor too close to condensate injection. The steam at the sensor has water droplets and an inconsistent temperature. The result is noisy measurement. A signal filter can help but the real solution is to move the sensor downstream where all of the liquid is vaporized and the cross sectional temperature profile is more constant. Just don't get carried away and locate the sensor so far downstream that the transportation delay exceeds 10 seconds at low flow.

(8) pH sensor too close to injection of reagent for inline pH control. Inline system have no back mixing and even the axial mixing is adversely affected at the discharge of the mixer. pH is an extremely sensitive measurement so small fluctuations in concentration will show up as high amplitude noise. Flashing reagents such as ammonia will require more time for the bubbles to dissolve. A signal filter can help but the real solution is to move the sensor downstream where the fluid is better mixed from turbulent flow and the cross sectional pH profile is more constant. Just don't get carried away and locate the sensor so far downstream that the transportation delay exceeds 10 seconds at low flow.

(9) Improper location of control tray in distillation column. Temperature is an inference of composition in a column. The proper tray is the one that shows the largest temperature change in both directions for an increase and decrease in the ratio of the manipulated flow to feed flow (e.g. reflux flow to feed flow). Process simulators can point to the best candidate. Temperature sensors should be installed in the recommended and adjacent trays and online tests done to confirm the best tray location. 

 

What are your comments?

You cannot post comments until you have logged in. Login Here.

Comments

  • Retrofitting a characterized trim in an oversized valve body can solve the problems described in most cases.  A characterized trim, for instance, can have high loss coefficient at low flow, and low loss coefficient at high flow, with gradual transition in characterization.  Many characterized valves have been proven successes nowadays in industry, for instance, in replacing two valves in parallel with one single characterized valve, such as the main boiler feedwater valve and the boiler start-up low load valve replaced by one single characterized valve.  On the contrary, the split range configuration replacing one valve with two parallel valves, which not only taking up space, but also increases maintenance burden long term while reducing controllability.

    Reply

RSS feed for comments on this page | RSS feed for all comments