Oil drops to $43 a barrel

Again, depends on the technology involved. A refinery built within the past 5 years - with fully modern digital control bus eg Foundation FieldBus - would be much easier to throttle or temporarily shutdown

If you have something with a much older 5-15 psi pneumatic control loop architecture, just bleeding the air out takes forever. That is, as you dump the pressure in the loops, condensation occurs.

That condensation must be properly adsorbed or drained, or extensive damage can be done to valve positioners, control loop servos, etc

I've ripped out older 5-15 psi loops back at the start of my career. It was kind of pretty how the small tubing was so carefully positioned with proper bend radii, etc

Then along comes Yours Truly with bolt cutters!

 

If your air is dry it's not such a big deal. The main problem I had with pneumatics was hoses cracking on the racks. Air leaks introduce a bias in the signal and take a while to track down.

For the critical valves we often had higher pressure air or nitrogen. One of my biggest gripes was valve positioners--they tended to get a lot of play in them and cause more trouble than they were worth.

We had a lot of digital on top of pneumatic and it worked pretty well. Our cracking plants and boilers did okay with it and hooked up to DCS control. And we had some other really nice proprietary stuff that manipulated the DCS...

 

Oh no! Not pneumatic controls. I had enough of those in the 1970s. Propagation delay was one of my personal pet peeves. It took nearly forever for a signal to travel a few thousand feet.

Tom

 

I don't remember having them on really fast loops. Most of the places we had them the runs were quite a bit shorter than that. I don't remember propagation delay being a problem, but I can see how they could be. When I had suspected propagation delay, it always turned out to be a leak somewhere or crimped tubing.

I have had some fun reducing sampling lag on boiler btu analyzers for off gas streams of highly variable composition. Part of the trick was to figure out how to maximize the flow from the closest sample point using the smallest tubing diameter. I think I also figured out a way to re-pipe the mix points so that they could be moved closer to the control room or else I would have been looking at the existing ~1,000 foot run...which wasn't working very well, obviously. Who wants a sample that tells you what was happening at the burner several minutes ago? If I remember right the savings basis was reducing the natural gas make up rate for steady minimum fire operation. It's been awhile...the sampling control work was just a side aspect of setting up better mix points and vapor-liquid seperator for additional streams. It was the only design I did on that old unit.

What I learned to hate was the ancient field mounted wizard boxes that really weren't controllers, just proportional. They were often used for level "control." I wasted more hours screwing with those things.

 

I had very little to do with actually maintaining pneumatic control. I usually ripped it out to put in DCS

Believe it or not, that problem still exists with modern digital buses and valve positioners. Not only play, but Sticky Friction - "stiction"

The device profiles now include software to partially exercise the valve and - with the built-in feedback - warn of excess play or stiction issues

DCS really wasn't that much better. Distributed Control System, except it really was central control. Lot's of individual discrete control wiring to make it all work

I like modern digital buses like Foundation FieldBus as the actual control loop can reside right at the loop level. Say you have a HX control, you can have the control resident in the valve positioner or the RTD.

Even if the conduit back to the control room is severed, the loop will continue to operate. You can also program in emergency shutdown routines on loss of central computer communication

 

Ah, the good ole days. Trust me on this: drink enough and those memories fade away

 

Yep, that had a lot more in common with plumbing/pipefitting than control engineering.

The old PID controllers should all be put on a gunnery range. You can then spend hours just shooting the s*** out of them

PID tuning was a nightmare with those older controllers. With modern "fuzzy" auto tune programs, like Emerson's EnTech and OvationTune, it's pretty easy

 

Start with a large derivative gain...

Tom

 

Ah yes, proportional, integral, and derivative. At least I had to actually use that college math back then ...

 

I am, at the moment, writing firmware for a small programmable controller: five PID loops, PWM output, with a bunch of optically isolated inputs and outputs. The whole thing runs on a microcontroller containing 128 KB of flash memory. I need to keep some of that for user programs, so I have to fit everything into about 80 KB.

The user interface is a small LCD and a keypad, with a multi-function rotary input. A USB connection is provided for computer interface.

Yet another squeeze job of programming.

Tom

 

The original Honeywell TDC 3000 had Multifunction Controller and Multifunction Controller II cards in the Hiway Gateway that were far more constrained than that

I'm going to guess motion control in your case?

 

I used to work with the Honeywell ("Funnywell") TDC 2000. We had a complex batch control process, and the chemical engineers decided to use the TDC 2000 for all of the loop control, but it didn't have good batch control capabilities. Allen Bradley PLCs were used for discrete control, but the TDC 2000 and PLCs didn't like to talk to each other. I had to glue the two systems together and provide supervisory batch control with a DEC PDP-11 system running RSX-11/M.

Ah, those were the good old days.

Tom

 

Oh gawd, TDC2000 predates TDC3000 like dinosaurs predate .... you get the idea

I'm curious how you were able to do batch control on a 2000? The 3000, you had the Application Module, Local Control Network, a History Module to store all that crap, and even then, you used DEB/PED to build the points, reconstituted the points, wrote the Control Language, made it resident on the Hiway Gateway MultiFunction Controller or Controller II, crossed your fingers, and turned it on

I still have some of my Control Language source code from a 1988 project. Looking back, it's a wonder that thing had enough processing power to draw the screen graphics *and* run the touch screen interface

As 2000 only had Hiway Gateway, it was too slow to do real time stuff. If you had a heavily loaded HG, forget about screen updates more often than once every 10 secs.

DEC PDP and Honeywell/Burroughs were common "real" minicomputer applications. Honeywell had a LCN/Token Ring adapter for the DEC. Somewhere I even have a brochure of this very device, a fantastically futuristic engineers office with his very own Engineers Minicomputer.

I always had the impression that given the cost, there were perhaps 30 such engineers offices worldwide. I do recall that in 1988, a slot 2 LCN card cost around $12,000 to replace

Don't even ask how I managed to fry it.

Hint: isolated grounding systems are incredibly dangerous. Never mind sparks, I had an actual arc event!

 

Two words:

Heavy

and

Drinking

 

ok u two

GET A ROOM!!!

 

Aw, you're just jealous