Control systems: really engineering? 1

[mrmojo]

Hi,

I have started to gain some experience in control systems in an industrial environment: programming plc's, creating wiring diagrams, cabinet drawings, selecting SCADA telemetry technologies.

Most of this stuff doesn't really seem like engineering, and in fact a lot of companies that do control systems consulting refer to themselves as "system integrators", which describes this type of work well i think: integrating control systems, instrumentation, and telemetry methods, and paying attention to things like communication protocols for different systems.

My previous experience has been in a different type of engineering, and I am sort of disappointed in control systems engineer: you don't really seem to need much technical knowledge and in fact a lot of job postings advertise for engineer or technologist.

In an industrial setting control systems seem to just be a relatively minor side show, the stars of the show seem to be process engineers or mechanical engineers.

Am I just being too picky, or missing something here? Is there a a more interesting aspect of control systems engineering that I have just not experienced?

 

[Skogsgurra]

Yes, surely there are more challenging aspects of control engineering. But very seldom in industrial control systems.

For many reasons (cost, reliability, availability, the need to pass knowledge about the systems to others and still some reasons) control systems of today are very standardized and need not much knowledge from the 'designers'. This shows. Many 'designers' of control systems are not even aware of the real properties of valves, relay coils, motors and other hardware components in a control system. Not to mention classical concepts like LaPlace, pole placement, Nyquist stability, Boolean algebra Kahlman filters and such things.

Good for me. It keeps me occupied with 'problem solving' that pays well and is nice and easy work. Often I bring my wife with me - especially if the 'problem' is in a part of the World that has nice weather or is interesting in some way or another.

You are so right in your observation. If you want challenges, go to someone that designs specialized control systems. Especially in the aerospace and new solar and wind power systems. There are new problems that need analysis and a deep understanding of design methods and trade-offs.

Good luck!

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottI2R]

Hmmm, how do I put this? It depends on what level you are working at. I mean, a "system" must be designed or engineered before it can be integrated. But, if you are just doing plug and play, I can understand your feelings.

the stars of the show seem to be process engineers or mechanical engineers.

Looking for the limelight are you? Perhaps controls isn't what you are cut out for. Not saying you are not good at it or anything like that, but maybe you just need more challenge. Only you can answer that. Maybe take your controls knowledge and use it to implement process improvements. That'll get ya in the light. Personally, I love working with plc's. Making a beer can crusher at home with one now. Real simple ladder. Like first semester stuff.

Well, I've rattled enough, good luck to you.

Scott



I really am a good egg, I'm just a little scrambled!

 

[byrdj]

I agree with Skogsgurra. I hope the industry continous to think that "controls" are not engineering. this would then continiue to provide Me (and the real Control Engineers) oppertunities to make a good living.

I'm currently on the sidelines of such an oppertunity where the multiply companies that sell control up grades seem to have forgotten about the equipment and the process they are trying to control and more interested in the complexity of thier HMI displays

 

[IRstuff]

Fundamentally, your question is no different than asking whether "washing machine engineering" is really engineering. There are lots of things that one can do on washing machines, including building one from scratch and possibly getting it to work right. However, there's lots of hidden facets to the design of such a machine, including:
> the hydraulics of pumping water in and getting it to agitate correctly
> agitating and mixing the clothes and wash water to get a thorough cleansing
> the fluid dynamics of spinning the basket to drain the water
> the mechanics of getting balanced spin with a cantilevered load
> the electrical and electronic controls

At any level, anything can be considered a system, and those that are simply using these things may not even be aware or think of them as "systems." Just consider the lowly resistors that are found in the feedback loops of analog control systems. You treat them as black(brown) boxes, yet, there is someone at Allen Bradley whose job it is to engineer the mechanics and materials of the resistor to make it a usable brown box.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[Skogsgurra]

No, IR. There is a big difference between connecting devices so they do their work and the design of a system like, for instance, a washing machine.

The control engineer in a typical paper mill or in a rolling mill doen't do much engineering today. They just get told what the function shall be and then make their equipment doing just that by setting software connectors and entering parameters. Very often without any insight why those parameters need to be set to the values they are told to set them at.

One example; a flying shear for very tough steel plate. Thickness from 4 mm to 16 mm. The whole algorithm for moving the shear block, do the shear work and open the shear again was delivered as a set of 32 bit unsigned integer numbers that no one could interpret or understand in engineering terms.

Same thing with winders for paper. Some manufacturers have open systems where you can see what different parameters do and some (most) manufacturers guard their settings like Crown Juwels and don't tell anyone what they are about.

Things have changed with the advent of computers in control systems. Before, analogue circuitry could usually be understood and one could - if one liked to - use classical analysis and synthesis methods to find why the equipment didn't behave or to make it behave even better.

That is usually not possible today, except for some primitive control loops.

Building a washing machine is still engineering, and your example is a good one. But being a GI instrumentation or control engineer in a typical process industry is not engineering any more.

I know the difference - and I feel it. I started designing drive systems at ASEA (now ABB) in Västerås. We worked tightly with the customer and every system was more or less 'custom design'. That was real engineering. What the guys do today is not. The result is better economy, reliability and availability. Which is good. But not a challenge for some engineers that may have misunderstood how indusry works today. They are very often misled by teachers that haven't seen an industry from inside for twenty or thirty years.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[itsmoked]

It still depends on your definition of control engineers. I consider myself a "controls engineer" but I generally design embedded control systems from scratch. Only occasionally do I reach for PLCs.

It's quite challenging. Often I have to know a great deal more about about the customer's system than they do.

"Control Engineering" has a lot of meanings. Sort of like "electrical engineer".

I know a "control engineer" who just designs rocket steering control systems. He's not buying this stuff off a shelf.

If you're looking for work these days, yes, you have to understand what the now vague term, "controls engineer", means to the company who's running the ad.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[mrmojo]

so in terms of control engineering as a commodity requiring little technical knowledge vs control engineering being more of a true engineering challenge where does control engineering for electrical utilities and the upstream oil business (particularly Alberta oil sands) stand?

it smoked: what kind of products do you do embedded systems for?

I have a graduate degree in physics, but only experience doing dumbed down control systems -- so any advice for getting someone who requires more customized control engineering to hire you? where do you even look?

 

[ScottyUK]

Control for the power generation industry can be pretty complex, for example turbine and boiler control is not a trival task by any means. Much of a control engineer's time will be spent trying to optimise performance, squeezing a little more power out for a little less fuel in. This could be anything from loop tuning to optimising actuator performance, improving instrumentation, to re-designing full control schemes. The chances to do the latter are fairly infrequent, but they do arise from time to time.

I'm currently working in oil & gas and although the process is basically quite simple on the surface there are a lot of details which make control quite challenging. Making it work is fairly straightforward - making it work well is harder.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[itsmoked]

itsmoked: what kind of products do you do embedded systems for?

In some cases I just design the electronics and in other cases everything, including the packaging, product shipping materials, and user's manuals. Here's a few I recall just now. These are controls/systems I've designed.

Bottling line glue dotters.

Hot melt glue machines.

Liquid chromatograph derivatizer controls.

Large building firepump controls.

Digital battery chargers.

Ethylene oxide sterilizer controllers.

Large stack flow sampling totalizers.

Oil field SCADA controllers.

A radio telemetry system for power plant commissioning load dump water column rise.

A 13 channel 20kW flat panel display process heater controllers.

A 48 channel vacuum chamber space simulation strut heater controller.

Temperature controller for for refrigerated rail cars, including electromechanical governor control.

Sauna controls.

LED fixture time-of-day dimmer controls.

Chiller controls.

Coffee and espresso machine controllers.

Refrigeration condenser fan-staging head pressure controls.

Smart 4-20mA loop controllers.

Fiber optics portable termination curing ovens.

Direct mounted solid state relay PID controllers.

Multi-stage thermostats.

Fiber optic wave division multiplexer temp controllers.

So a 'controls engineer' can actually design stuff and not just integrate stuff. I've done my share of integrations and frankly.. they can be a lot of fun too.





Keith Cress
kcress -

http://www.flaminsystems.com

 

[mrmojo]

itsmoked/keith:

Those are some pretty cool looking projects on your webpage.

At the risk of exposing my technical ignorance, and if you don't mind me asking:

couldn't most of those projects have been done using a cheap pc running labview connected to the appropriate industrial display/controls?

i asked people where i work why the industry uses plc's instead of controlling industrial processes directly form pc's and was told that it's largely a legacy/support issue.

however for one off projects like you're doing wouldn't programming a cheap pc for each individual application be easier than coming up with unique electronics?

 

[Skogsgurra]

Since Keith is in the Bay Area and hopefully gets a well deserved nap, I (in Europe) will anser in his place.

Many systems can not use a PC simply because it may be a critical system that cannot tolerate the PC hickup that was so common in the earlier days of PC - and still is, I would say.

Also, the restart after power downs or brownouts is also something that may be a problem with a PC.

There is also the space aspect and power consumption.

It may be counter-intuitive, but a small project with a custom designed board and custom software is very often better and cheaper than a PC and, for instance, an NI box of cards. Also, the LabView programming 'language' isn't that great, after all.

Keith prefers custom designed boards (and so do I). A typical control board will not take more than a day or two to design and the board guys will produce ten boards in less than a week if you are in a hurry and at a cost that is less than the tax paid for an NI board. Custom boards fit space and form factor needs and do exactly what you need them to do - not 'almost' or 'not quite'.

Programming is also an easy business if you have done similar projects before.

The real costs are usually in understanding what the customer really wants and also to make him understand what is possible and what is not. That problem is the same whatever hardware or programming system you are using.

When it comes to more than ten systems, there is no discussion.

Did I forget anything, Keith?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

I wouldn't trust any critical control to a PC. Simple as that.

Maybe you could convince me to run something important on a Sun SPARC platform which historically are very stable and can run without a reboot for years at a time, but then cost becomes prohibitive unless you can use the processing power for something else.

PLCs are reliable and maintainable, and have long-term support. I can run an application developed for an SLC 5/01 processor twenty-odd years ago and know it will run on an SLC processor I buy today. As for a PC application from 20 years ago, well... good luck!



----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[ScottyUK]

I also meant to say that embedded control and PLCs have, until recently, been considered pretty much immune to viruses and malware. All that changed with 'Stuxnet' which is designed to infect Siemens PLCs - using a PC as the means of transmission - but PLCs and embedded controllers are still a comparatively low risk compared to using a PC.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[MacGyverS2000]

Scotty,

I'd say that's because the PLCs these days pretty much have standard OSs running inside of them, and therefore are no different than a PC in terms of virus issues. My idea of a PLC (or embedded system) does not include Windows/Linux anywhere in the description. ;-)


Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

And that's only because those particular embedded processors are running relatively tiny OSs that don't always provide much in the way of infrastructure.

Larger embedded systems have to support GigE, file systems multitasking/multithreading, and the OSs associated with those applications such as VXworks are not immune from bugs. Additionally, the control applications that run on such systems have thousands of lines of C code, and the verification/validation of such code is not always robust.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[controlsdude]

I guess anyone can put a puzzle together is what the poster is saying.

But really the game is can you do it within budget, on time and at the rates stated for what the machine was designed for. Some projects are more challenging than others. Make the prints, write the programs, and debug and startup the equipment.

Some puzzles have more pieces than others, and some puzzles have pieces that dont fit exactly. A controls engineer does operate on this area "of puzzles" to me.

I guess my question is how many pieces does your puzzle have? Is it a 10 piece or a 10000 piece?

 

[itsmoked]

Nice description Dude.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[BrianE22]

If you like the math, look into companies that design servo system controllers. I'm amazed at what the CNC machine tools in our shop can do. There has to be some nice engineering going on in the hardware and software for those controls.