We have an application where we are controlling the loading of a Press via hydraulic cylinders fed by linear hydraulic servo valves. The Bosch valves are controlled by analog-only Bosch controller cards which do the closed-loop pressure control. These cards receive a supervisory pressure setpoint from our DCS (Honeywell TPS).

We would like to do the closed-loop control directly in our DCS. This would greatly simplify the design (eliminate several troublesome signal transducers as well as the Bosch controller cards). However, the OEM is cautioning us that this probably won't work because the servo valves are high speed (operate full stroke in anywhere between 10ms to 32ms) whereas our DCS has a relatively slow (250ms or 500ms) scan rate. The OEM says warns that our tuning will have to be so slow that it will take forever (> 5 minutes) for the controller to reach desired setpoint.

I'm not a control expert, and I can't find any good literature or articles addressing this. Has anyone successfully used a traditional DCS for closed-loop pressure control where fast-acting hydraulic servo valves are the final control element?

 

Your OEM supplier is correct. The digital update rate must be much faster than the process. One simple way to look at it would be that sampling at 4 Hz, the highest theoretical signal that can be encoded is 2 Hz. Typically the servo update rate must be at least ten time the systems closed loop bandwidth.

 

Hi,

Your OEM is partially correct. You *can* control a very fast final control element and your performance *will* be limited by the scan time of the DCS controller.

The servo valve will be seen by the controller to provide an instant change in pressure so that in itself is not a problem. However, the controller scan time will appear as a deadtime inside a very fast control loop, which makes achieving a fast response difficult. But you could control the pressure but with a slower closed loop response. However, I suspect you wouldn't have to go as slow as the 5 minute response time suggested, 10 seconds would probably be achievable with a 0.5sec scan time. Ideally you should do some analysis to get more confidence about what is achievable.

Hope this is useful,


Andy Clegg_______________________________________ [email protected] __

Applied Control Technology Technology Consortium
50 George Street, Glasgow, G1 1QE Tel: (+44) 0141 553 1111
http://www.isc-ltd.com/actclub Fax: (+44) 0141 553 1232
______________________________________________________________________

 

If you close the loop through the DCS and it can only service the loop every half second, your pressure control bandwidth will be only about 0.2Hz considering the 10:1 rule of thumb. This means it will have a time constant of about 1 second. Therefore if the press encounters a hard spot or other load disturbance, it will take about 3 seconds (3 time constants to correct by 95%)to respond to
the disturbance. That sounds pretty bad to me.

Tom

Thomas B. Bullock, President
Bull's Eye Research, Inc.
N7614 State Road 149
Fond du Lac, WI 54935-9507
Ph: 920: 929-6544
Fax: 920: 929-9344
E-mail: [email protected]
www.bullseyenet.com

 

The basic rule-of-thumb is that the controller should have an update rate 5 to 10 times faster than the loop response time. Just because the servo valves can operate in 10mS does not mean that the loop response time is also 10mS. The loop response time is the time it takes between changing a setpoint and seeing a change in the process variable.

warren
http://www.pc-pid.com

 

I certainly haven't tried to use a DCS for motion control of any sort, but can say with some confidence that I would never try to run a servo loop (even hydraulics) at anything less than a MINIMUM of an order of magnitude faster than you are stating for your DCS. I have run servo hydraulic control using high bandwidth servo valves, and our controllers default to 2.2 kHz on the servo loop. I have slowed that down for hydraulics at times due to the slower response, but never gotten below 220 Hz.

Davis Gentry
Delta Tau

 

Hi,

The question I have is more concerned with what is the frustration, rather than a solution that is anxiety free. Yes, closed loop hydraulic servo systems are very fast. Yes they can do a lot of damage in a hurry. Can you eliminate the cards, the sensors, et al. You sure can. But it depends on your process.

In our industry, a lot of the hydraulic stuff is being replaced by electric servo and stepper motors. Very complex. Doesn't leak. But it still has encoders plus.

So is the frustration troubleshooting, or process variance, or maintenance, or what.

We use closed loop servo hydraulics for parison control for blowing bottles. The tooling is designed so that if the system does screw up, with or without human assistance, and the tooling goes flying to end of stroke inadvertenly, there is nothing hurt but pride.

Those systems are supervised by PLC's that issue a command signal roughly every 12 ms, that command signal is fed to an analog error amplifier, then that result to a servo amplifier. This simple system uses one LVDT as feedback.
But that analog error amplifier can easily operate at 10 khz or higher.

So be careful that the tiger that you have doesn't get changed for one with bigger teeth.

Old John

 

The discussion on hydraulic servo valve control has been interesting. However all the discussion revolves around the digital control capabilities in controlling an analog servo valve. There are other considerations in applying hydraulic servo valves to industrial applications.

Both hydraulic cylinders and rotary hydraulic servo actuators have performance limited by the oil under compression (trapped oil). For cylinders the length of stroke determines what the trapped oil under compression will be, and this determines what the hydraulic resonance will be. One figure of merit limits the resonance to be no lower than 200 rad/sec for a stable servo.
Hydraulic resonance= square root(2B Dm^2/Jt Ve)
where: B= compressibility= lb/in^2
Dm= motor displacement= in^3/rad
Jt= Total inertia at the motor= lb-in-sec^2
Ve= volume of oil under compression= in^3

For rotary actuators where the servo valve is directly connected to the motor through a manifold, it is possible to get about 60Hz or more with no load on the motor shaft. Most hydraulic servo motors can not stand any side loading on the bearings in the shaft end of the motor. Therefore some coupling device such as a Thomas couplings is required to minimize misalignment of the motor shaft and a load shaft. These couplings have inertias much larger than the motor shaft inertia. This requirement along with the reflected machine load inertia connected to the motor also lowers the hydraulic resonance. The result of these added loads reduces the hydraulic servo bandwidth on industrial machines to about 15 to 20 Hz.

In general, if a servo drive has a very large torque requirement where electric servos can not provide sufficient torque, the hydraulic servo certainly will. The drive sizing criteria for hydraulic servos is getting the hydraulic resonance high enough. The servo valve is a 2-stage device. The output stage is the positioning of the motor spool. The input stage is an electric torque motor that has a very high bandwidth and an impedance that must be matched in the servo valve controller. The servo valve amplifier can be designed in-house if you know what you are doing or the suppliers can provide this amplifier.

 

The manufacturer is correct. You would have stability issues. What you need is a fast update rate for the loop closure. You could also use an Analog loop closure board. We typically use Delta Computer (Battelfield , WA) for the digital servo hydraulic control and Moog for the Servo valves for high response servo proportional valves.

Best regards,
Don Dougall