Need Help With Active Pressure Control 2

[dfowler]

I am attempting to pressurize a chamber to pre-determined pressures for varying lengths of time. I am using a fan that can output 2500cfm at 14.5”SP. The fan is connected to a VFD. I am pressure testing against a garage door to simulate wind loads so I anticipate that there will be potentially a great amount of leakage across the door. It is not known to what extent this leakage will be and most likely it will increase as the time of the test progresses. Therefore I will need to dynamically adjust for this increased leakage. This can be done by turning up the VFD connected to the fan and delivering more cfm and pressure, but this most likely has too slow of a response to compensate.
I want to employ a valve that is to be open during the test. This valve can then modulate its orifice to allow lesser flow to escape (as more does across the door) and therefore hold pressure within the chamber. I know I could build a system myself with pressure transducers, a PLC, and a valve with an actuator - but it might have too slow of a response. I am wondering if there are any “off the shelf” products that would suit my needs. I have located one integrated smart throttle valve from MKS pressure measurement and control products that seems would work for me.
I intend to dial up the VFD to a calculated target frequency that should deliver flow somewhat above the desired pressure. The excess flow (and pressure) will then bleed off through a combination of the leaks in the door and through the exit valve orifice. If the pressure is sensed less than the desired amount the valve will close more and allow less flow to escape. If the pressure is sensed greater than the desired amount the valve will open more and allow more flow to escape. This needs to be a fast response action as it would not be favorable to have high amplitude on my pressure variations.
Does anyone have any experience with fast response active control valves? These are low pressures (less than .5 psi), but relatively large flow rates for a valve with such minimal pressure. I will probably need to bleed off between 400 and 1000 cfm most of the time. I think I need a valve size of at least 6” diameter. Thanks.

Davey

 

[imok2]

Dfowler , If you use a 6" butter fly valve s and want to dump 400 to 1000 CFM the velocity would be 2000 to 5ooo ft/min pretty loud, however you could parallel 2 with a manifold. Control should be pretty simple with pneumatic

 

[dfowler]

Good point imok2, I hadn't thought of the noise potential yet. Would you suggest pneumatic control vs. electric stepper motors? I would like variable control with my valves, not just on/off. Also I would love an integrated system so that I do not have to calibrate. Anyone have any suggestions on components? Thanks.

Davey

 

[imok2]

Dan, I like pneumatic because you have constant proportional control ( 0 to 100%)and it's easy to understand,install and control. Hope this helps. Stay tuned you may get more input from others!

 

[ChasBean1]

Davey, your initial net open leakage area from the door would be about .25 ft2, similar to a ~6.8 inch diameter hole (given 2500 cfm, 14.5 in. w.c. static pressure). This is based on an ASHRAE leakage estimate.

Have you considered making the whole test human + stopwatch? Why do you need actuators? Consider four ducts side-by-side off the pressurized chamber, size 2", 3", 4", and 5" round, with low-leakage butterfly type air valves on each. You could get numerous pressures with this configuration instantly, just by stroking the valves manually.

If not, the same configuration with fast-acting two-position dampers could be used. Just some thoughts. Good luck, -CB

 

[dfowler]

Thanks Chas, but I am lazy. I want to set this all up and not have to think anymore - just push buttons
Just kidding... I have thought of that very idea somewhat. My latest plan is to get a VFD for fan control and integrate a pressure transducer with it. With the use of software hopefully I can download new setpoints to the VFD and have the fan hold steady when each desired pressure setpoint is sensed. We can manually count down the time required to hold the pressure and stop the system ourselves.
With a manual (or electric) butterfly or gate valve in the system we can start up with this valve open. The pressure transducer will not sense enough pressure and will ramp the fan up to full speed. We will then close the control valve and let the pressure transducer signal the fan to ramp down in speed to hold the appropriate pressure setpoint.
Now I just need a good PLC and pressure transducer combination. Anyone have suggestions for a configurable software and PLC package for a 3 phase 12 hp motor? I will post this question in the PLC group as well. Thanks.

Davey

 

[quark]

You need not go for a PLC in conjunction with a pressure transducer and a VFD. Almost all VFDs now used for HVAC application can directly take 4-20mA input (for ex. Danfoss and ABB etc.) and this can save your precious response time (in milliseconds)

Go for a good pressure transmitter and you are done.

But as suggested by imok, pneumatic control is better for it's fast response when compared to electrical control (generally a difference of 2 seconds)

However, I like CB's idea of manual control.

Regards,

 

[dfowler]

Thanks for that info quark. I will want to change my setpoint often however as we need to test for 50%, 100%, and 150% of the design load (i.e. pressure setpoint). The design load will vary from test to test as well so there will be a very high number of different loads we will want to try.
The human factor would be great for ease but this test has to have calibrated instruments and to be certified by an independant PE - they might not like that We need to record data (force and deflection over time). The force has to be held steady against the door for 'X' amount of time and the door has to regain at least 75% of its maximum deflection during the test (if it bent outward 4 inches it must relax back 3 of them after the test).
I would like a PC based control system where I could use analog signal collector cards to gather data from deflection sensors and a pressure transducer. I just need to record deflection and pressure vs. time.
I would also like to use the PC to download new fan speed/pressure control setpoints to the VFD for as this would have to be set often and would probably be a pain with the VFD's control panel.
Does anyone have any experience with PC based controls or data aquisition? Thanks for any help.

Davey

 

[quark]

That is not much difficult, but generally available PC based control systems will cost you a fortune as your system is with only 2 AIs(pressure and deflection and 4 AOs (pressure, deflection, time and fan speed)

Windows OS has something called Hyperterminal which can capture data directly from an RS232 port. You just need to build a driver for the interaction and this will be the cheaper solution. I saw such applications, particulary for data capturing.

Regards,

 

[ChasBean1]

A chamber pressure transducer controlling a fan VFD could kill the need for the bypass assemblies. My experience with ABB drives is that you can change inputs instantly but the fan would not instantly respond. E.g., a step change from 15 Hz to 60 Hz would immediately display the change on the upper left corner of the panel, but then the fan would take about 30-45 seconds to execute the change. There's usually a time delay although it's probably adjustable.

I think you've pretty clearly defined the problem - the next step would be to create specific RFPs. Hire a reputable mechanical contractor that has the right contacts for drives, sheet metal, and ATC (Johnson, Siemens, Invensys, etc.). This is not rocket science from a controls standpoint, so long as they clearly understand the scope and that the result of the execution actually represents the intent... This could result in an iterative process (use a 3rd party commissioning agent) for which the controls contractor would require contingency.

You could achieve the result with your own work (which I think is always better). But this leaves you with the burden of knowing and handling all facets, such as programming interfaces, etc. Things you learn on the fly have already been done by others. Put yourself in the position of engineer (brain) behind the project and hire (and work closely with) a team to get it done.

Best regards, -CB

 

[ChasBean1]

As a side note, I'd be concerned about the 2,500 cfm fan size. It sounds a bit small for this application...

 

[dfowler]

Ok, here is the latest control scheme:
We have a VFD with a closed PID loop and an integrated pressure transducer. We can easily change the setpoint in between tests via the local control pad on the VFD. Each new setpoint will correspond to a known pressure value. The VFD will run the fan up to the signal set point and hold steady. I intend to set any delay in frequency change response as low as possible, however the fan speed should not have to change much during the test. The fan might have to ramp up slightly to compensate for increased leakage during the test. The 50%, 100%, and 150% design pressure tests will be done approximately 5 minutes apart giving the door time to relax in between as well as allowing time to change the setpoint.
We have a large manual exit valve that will be open at the start of each section of the test. This valve will instantly exhaust any pressure before it builds in the chamber. The pressure transducer will thusly not sense enough pressure and will signal the VFD to ramp the fan up to full speed. We will then slowly close the exit valve. As the pressure in the chamber builds, the pressure transducer will signal the fan to ramp down in speed and the pressure will be held at or slightly above the appropriate setpoint.
The pressure transducer signal will also be fed to a DAQ card in a PC. With monitoring software we can begin a timer count when the pressure setpoint has been reached. Once the appropriate time has expired, the DAQ card will output a signal to a relay that will cut the power to the fan and will simultaneously open an additional exhaust gate for instant pressure relief.
We will also be measuring deflection over a time constant with string potentiometers during the test. These signals will be collected and logged into the PC by the DAQ card and software.
Please let me know if anyone finds any holes in my control logic or sees an easier way to my end

Davey

 

[ChasBean1]

Davey,

I guess you are confident in the fan size.

You indicate three devices will affect your primary parameter, which is chamber pressure: 1) Fan VFD, 2) a manual exit valve that will open at the start of each test, and 3) an additional exhaust gate for instant pressure relief.

What I wonder is that with 2) and 3), why have 1), or with 1) and 2), why have 3)? You noted that the fan speed (1) should not have to vary much during the test, except to account for leakage. I really don't get this. Adjustment of (2), the modulating damper, IS leakage. If the VFD compensates for leakage, then the pressure remains constant and you don't have the varying test pressures you are trying to achieve.

You're also at the mercy of timing. Shutting (2) the control damper will cause an instant pressure change for which a VFD adjustment will likely be too slow to compensate.

It sounds like you are knowledgeable and on the verge of a solution, but I'm just a little miffed. I think you are over-controlling to achieve the objective.

 

[dfowler]

Thanks Chas,

I think I am confident about the fan size. The fans will mount directly to the chamber so there will be no piping loss. The pressure is what I ultimately need. If I've sized the cfms too low then it will take longer to fill up to pressure.

A sample test criteria set may be:
We want a door rated for 100mph. According to our charts (from the IBC) 26psf of pressure would simulate 100mph wind. We have to load the door with 50% of the design load (13psf) for 10 seconds. We then stop the test for 5 minutes to let the door relax. We then apply the design load (26psf) to the door for 1 minute. We then stop the test for 5 minutes to let the door relax. We then load the door with 150% of the design load - the test load (39psf) for 10 seconds. We then stop the test for 5 minutes to let the door relax. If the door is operable through its full range of motion after all of this it passes.

The test load is the real deal breaker here after the first two loadings. We want to minimize the time for this part as much as possible. That includes ramp up, load time, and ramp down time. The 10 seconds only applies to the load portion so if it takes 10 seconds to ramp up and an additional 10 seconds to ramp down we are stressing this door unnecessarily.

The fan VFD will control the fan at a set pressure point for us. You are right in so far as the fan VFD on its on closed PID loop eliminates the potential need for the manual butterfly valve. I am just afraid of overshooting the pressure point and am visualizing bringing the fan up to full speed with the manual valve wide open. This will allow the air to escape while the fan gets up to speed and the chamber will not pressurize yet. I will then slowly close the manual valve; by the time it is closed the VFD will have gently taken over at the set pressure point without overpressurizing. At least this is my hope and aspiration anyways

The exhaust gate I want to be controllable electrically. I want it to be closed with a power source and fail open without power. I can then have a stop button to kill power to the entire system and instantly depressurize the chamber.

I also want the system to be neatly automated as well. I have analog signal output capability from the DAQ card in the PC. A counter will start when the PC senses the chamber at or slightly above the set pressure point. When the correct amount of time elapses a signal will sent from the DAQ to a relay that will cut the power both to the fan and the exhaust valve simultaneously. This will exhaust the pressure from my door instantly.

Especially during the final leg of the test (the 150% of the design load portion) I want to minimize the time the door is under force to exactly the required 10 seconds. Me or one of my cohorts with a stopwatch could easily run over by several seconds which is 30-40% too much time. Also just cutting the fan power alone might not be quick enough in relieving the pressure from the chamber.

Your statement about overcontrol is well taken. I may hold off on #2, the manual valve, and just let the VFD ramp up and see how it reacts to its PID loop setpoint. I can always put the manual valve on my chamber later. I am just a little afraid because the fan will push up to 15"WC and many (most) of my tests will require only 2-5"WC pressure. I don't want it to go barreling up to speed and go overpressure for any length of time if possible.

Sorry for being so wordy
Davey

 

[ChasBean1]

No problem, I think I got most of that. One thing you noted (way up top) - "If I've sized the cfms too low then it will take longer to fill up to pressure" is not the case. There's no filling up to do - room pressure will be at nearly instant equilibrium. We're almost on a negligible scale in this application (not like we're filling a tank to 2,000 psi). If the fan is instantly at speed X, the room pressure will right away be at equilibrium based on that speed and the amount of room leakage. You might see a bit of a lag if you did this in a shopping mall (maybe)... you'd have to get into how much more dense is the air at 15.2 psia, 15 in. w.c. higher than atmospheric, 14.7 psia. We don't have to go there, just trying to make a point.

I think you're on the right track. I would stay on the track of simplifying. Instead of any blowoff damper you could just kill the fan (I realize there's coast-down, not sure if that would be acceptable).

 

[dfowler]

Wow, that is simple. You make a good point though. The manual valve was only to improve a (possibly) slow ramp up to pressure time. The exhaust gate is intended to improve a (possibly) slow ramp down time. If I am going to wait to see what my system response time is on the ramp up, why not wait to see what it is on the ramp down as well before implementing either the gate or the valve?

This will eliminate the need for any external relays to cut power as I can just use relay the ouput on the DAQ card to signal the stop command on the VFD when it is time.

Thanks, I think I will go down that road first and see how it works out.

Davey