Factory Energy Management

[Lancie1]

Mike,

You got it, design & build on the fast track! Bids for the WHOLE thing must be in in 2 weeks. What am I doing here? I have got to go and get back to WORK!

 

[elevmike]

Actually it's Design as you are Building. Big in the Casino busniess, that's all they do (as it seems)...Get ready emotionally for the big sprint. It takes a special cast of characters to pull them off and have everything fall into place. Make sure all the contractors have cell phones (field guys too)and email. Good communications is a must! The job Super needs to be a fast thinker with a cool head. He's the key to winning or losing. Lots of impromptu meetings. Pick all the subs based on past performance, NOT just cost.

BTW, Budget?? What Budget?? There's no Budget!!

 

[Sheldn]

The method of saying "give me a computer" is typical of Asia and what's amazing they get pretty much what they want. For North America it is not what is done in the design-bid-build market. The design-build market does kind of do this but there has to be some parameters. How can the customer expect to get what they think is an "energy management system" when every person has a different intrepretation. Do I get graphics? Data Logging? Does the supporting network come with it? Does it help me save money?

If you know how to play the game, you should provide them a cheap laptop that allows them to connect to each manufacturers equipment through a serial port. Give them the serial cable and the computer, then show them how to use hyper terminal to access the command line display to monitor and program their system. It could meet the spec. Then when they start explaining what they really want, give them a price to do it.

 

[darrenj]

Lancie1..Yep sounds like design build to me..

My firm specializes in this..You basically become the engineer..If it doesn't work at the end you fix it at no cost....This seem like a major risk to most people but it isn't as bad as they think..let me give you some examples..(They are electrical but you will get the point..)\

An engineer will assignee 4 circuits for a "Pod" of 8 workstations (You know the snap together office furniture..cubicle farms) They will also install one bell/horn for fire alarm per office..they ask for a dedicated circuit where ever there is any doubt that a fax or printer will go..

Now in reality and from experience i can tell you that you only need 2 circuits for those 8 workstations..($ stations to a circuit) If its bells then strategically placing the bells can reduce the numbers drastically for fire alarm and most faxes and printers don't require a dedicated circuit anymore..so why do the engineers do this?? CYA..IF There is a problem after then the engineers have to pay us to go in and fix it..believe me when an engineer screws up we make them pay!! however if i design / build and there is a breaker tripping i send a guy in and an hour later its fixed..cost me what $100.00? or so..

Now the key is to estimate it right..I know that on 8 workstations 2 circuits is good..but i price 3 or 4..i look at the offices and find out they are not insulated or have a glass side light..i then look at the layout and place the bells to where i figure they should be good..then add a couple or so depending on the size of the project..keep doing this and trust me you are covered for any Little mistakes that come up..The client is happy because you don't biker over every single change..(And sometimes the change works to your benifit)..Your happy because they are happy and your making a good buck..

I just finished a fast track 35,000 sq foot office space..The clients were looking to save money so we chopped away at the engineer drawing and gave them our number..To cut a long story short the client saved over $100,000 , we made a damn good buck..(About 30% return)..we reduced the fire alarm bells by about 20 during construction..(about $100 pop) and had to add in 4 during inspection to meet DB levels..

With all that being said..if you are not used to this way then it can be a hell of a ride..stress goes way up because there is no one to point the finger if it doesn't work..just the client saying "I don't care just fix it"

I have only lost my shirt on one job..it was a restaurant and i will NEVER NEVER do another one..


To Add

As sheldn said..there is at least some guidlines spelled out..and our quotation is quite expiciate..for example..we shall reduce the number of bells to 17 However if additional bells are required by the city they shall be installed at a cost of Blah |Blah Blah

This usualy ony stands for big ticket items..the odd plug here and there gets desided on the fly..and depending on site conditions...example.."Do we need a receptical here in this hallway Bob?"..said bill looking around..Bob says.."Well look there is one about 10' away..so no dont put it in.."

 

[Tripper]

I have read some negative comments about BACNET and BACTALK. Does anyone have any other comments about those HVAC control systems?

http://www.bacnet.org/Bibliography/DMF-7-96/DMF-7-96.htm

 

[BobB]

That looks to be a very interesting article. Will read it properly when I have time.

BACnet has caused me quite a few problems quite frankly. If you are heading this way, be very careful. Make absolutely certain that the modules have been tested for compliance in the BACnet laboratories. I do not believe they are testing software and drivers yet but could be wrong.

Had a driver in Citect SCADA (cost an arm and a leg) talking to a BACnet system. Reading values, alarms, information and changing temperature and humidity set points.

1) The first driver was slow and unreliable.
2) The new driver I received was installed and worked a lot better BUT fans kept turning on and off. Thanfully it was not the chillers.
3) The BACnet module system manufacturer kept blaming the driver.
4) The driver manufacturer was very helpful and quietly spoke to the module manufacturer at an ASHRAE meeting one night and both got together to help resolve the issue.
5) It turned out that the BACnet modules were not fully compliant. The new driver used a call that the module manufacturer had not tested.
6) The module manufacturer had to replace the whole system with fully tested and certified modules and all was well. Installed the new driver and no massive problems since. Jus a lot of annoying things like communicvations dropping out between the driver and the modules. THIS TOOK 14 MONTHS of frustration and the module manufacturer making out the driver was at fault and that I was an idiot and did not know what I was doing. Basically, I had to help sort out everyone elses problems. They were certainly not mine. Cost me about $14.000 AU over the 14 months. No offer of compensation of course.

From the same Citect SCADA system we communicate with 5 Omron CS1 PLCs - these have a total of 11 Device Net loops running off them, and communications to a Wormald fire system via Modbus RTU.

Quite frankly, the only bits hanging off the SCADA system that gave me absolutely no problems were the Omron PLCs with the Device Net I/O hanging off them. This included explicit messaging to AB Powermintor 3000 devices and a lot of Omron digital I/O. The SCADA also behaved faultlessly.

The nasty with the Wormald fire system was that the RS232 port was only a pseudo port. The voltages were all wrong. Used non-isolated RS232/485 converters and watched chips explode off the PC board. Learned an expensive lesson and now I only use Adam 4520 isolated converters. The only problem here is that the Adam units require 24VDC to power them. Luckily these are normally used in generator systems and everything runs on 24VDC.

LON experience has not been so good - the article alludes to the fact that it is a small network.

The other thing I find interesting in the article is the reference to the fact that neither of these protocols is deterministic. My experience with BACnet is that it is as slow as a wet week, probably due to this. Quite often takes 10-12 minutes to pull a couple of hundred tags into the BACnet driver. It is an OPC driver and the exchange of data with the SCADA system is actually quite quick. Just have to wait a hell of a long time for the BACnet drivers and modules to get their act together.

My impression of BACnet is that it is an immature technology and that the industry really has not got it's act together. It may be OK from controller to controller but access to outside appears to have a long way to go.

Have never been impressed with LON and stay away from it like the plague wherever possible.

Both technologies appear to be quite expensive also.

Give me a PLC system with a deterministic network (this knocks out Ethernet and Ethernet IP in my book) anyday. I am used to seeing things work seamlessly with my preferred PLC network - just set up transfer words in a driver and it all works automatically and seamlessly and is FAST. Have one system running at 2 meg over a twisted pair and Citect kernel reports up to 15,000 digital reads per second and about 150 analogue reads per second (it is extracting data that is not mapped on the network but is being extracted from each of nine PLCs) - this does not include mapped data that is automatically being transferred between PLCs peer to peer. The mapped data represents about ten channels of digital words and 100 data/analogue words per PLC times nine. Quite a lot of data. The scan time on the network rarely exceeds 26 milliseconds.

Horses for courses really but be very careful on your selection of equipment. The HVAC industry really does not normally require much speed because things happen very slowly. A bit different with machine control. Fast networks are usually an advantage.

 

[Jiri Toman]

Energy management systems are supposed to optimize your system for minimum energy usage. For example in a warehousing application you would run your conveyors only when needed. On an injection molding machine you would only run a feed hopper when machine is on etc.
One system that has been touted to save money is an AC drive or a soft start with energy management built in. Usually AC induction motors are oversized to cover the worst case load scenarion.
For example on a crusher the load varies wildly. This would an ideal candidate for an energy management type AC drive, however
air compressors are not! Allen Bradley has energy management built into some of their soft starts.
I doubt if any AC motor on an HVAC system is a good candidate for
an energy management type drive. You will need to look for innovative ways to switch off what is not needed. For example
if the room is not occupied you make sure that lights are off
(you will need to use lighting network such as DALI or similar)
and temperature is at optimal energy efficient value (low in the winter, high in the summer).

 

[Tom Jenkins]

It's been a while since I've done this. A lot of the variability comes in the type of plant it is. A discrete manufacturing facility isn't going to be much different from normal building management and HVAC control. In a process facility the real savings may be in optimizing the process equipment, which required a lot more expertise. Just a couple of thngs from the mouldy recesses of my brain:

Hardware cost comes down to number of I/O points and basic architecture. I like to use distributed architecture, with a lot of small PLCs linked by a comm network. If you know how many heaters, lights, water pumps, chillers etc. you can guess that pretty close.

The real money is in the programming. Include a pretty good list of control loops, controlled equipment, and strategies in your proposal.

Most common strategies include optimized start stop, load (demand) shedding of non-critical equipment, and night set back thermostats.

Variable air volume for the HVAC control is a big potential savings, but it requires specific hardware. Look at the equipment specified for HVAC for guidance.

Take a look at the ASHRAE for some texts and guidance.

 

[Jiri Toman]

I like to use distributed architecture, with a lot of small PLCs linked by a comm network. If you know how many heaters, lights, water pumps, chillers etc. you can guess that pretty close.

PLC manufacturer's have been trying to make inroads into the building management and I don't doubt that you can control
buildings with PLC's, however from my experience I can tell you that having a dedicated building management system such as Andover will make things easier. Some of the things that Tom has listed have already been built into these systems. If you have never done a building management system then having a specialized controller will save you lot of headaches.

 

[Tripper]

My experience with BACnet is that it is as slow as a wet week, probably due to this. Quite often takes 10-12 minutes to pull a couple of hundred tags into the BACnet driver. It is an OPC driver and the exchange of data with the SCADA system is actually quite quick. Just have to wait a hell of a long time for the BACnet drivers and modules to get their act together.

I dont doubt your experience with Bacnet Bob, although my experience is quite the reverse.
The key to bacnet is how the information is passed and a properly designed netwok is very quick and reliable.
Where possible controllers need to have the relevant points local, ie to be able to operate stand alone.
Where points have to be passed, they need to be prioritised and setup accordingly using the Pri 1 to 10 priciple and using the Do_every statement and COV (change of value)only for critical points.
The last building I did we had approx 80 points (from different panels) per graphic screen (VAVs)and a refresh rate of 1S was possible, although you dont set it up that way because after all they were only space temperatures.
The Rs485 network ran at 17K baud (with 90-100 controllers on each sub net) and the main ethernet bus with 30 system controllers.
I think I'm suggesting for a 10-12 min for a couple of hundred tags somthing is seriously wrong; poorly programmed, designed or bad installation, after all this is oz and there is not the training avaliable on DDC systems as to PLCs.

 

[BobB]

I think I'm suggesting for a 10-12 min for a couple of hundred tags somthing is seriously wrong; poorly programmed, designed or bad installation,

This was done by the BMS company with their own equipment. I believe their update times were OK, with the exception of an RS232 serial interface to AC units in the data rooms, switch room etc. I forget what you call them - they are the box type ones usually used in data centres. Why would you use RS232 serial interface to 40 odd of these things and expect the comms to work OK? RS485 or Ethernet would be OK. The system spends half it's time reporting comms errors. Turn a unit off and about 8 of them go out. Stupid!!! That was really bad design on their part.

I believe the fact that the so called SCADA engine I had to hook into was a local product may have a lot to do with it. It was very slow. It also had to handle the very badly designed comms network to the AC units. When I installed the latter driver they had to change the code for the SCADA engine as it did not even have an "online" bit and the new driver would not recognise it without the bit on. Amongst a lot of other discrepancies that I have not mentioned. As I said, I have only mentioned the tip of the iceberg, it was quite a dramatic project. The consultant wanted the BMS done by our company in the PLC system but the mechanical services contractor refused to warrant the mechanical services if the job was not done by his preferred BMS contractor. Not a good place to start. Will have to get together one day and chew the fat.

If you have used BACnet in Ozz, you have probably worked for the company I speak about. They are supoosed to be the "doyens" of BACnet but had non compliant controllers and had to replace them all. The BACnet labs are testing for compliance now but I have not heard that they have progressed to a point where they are testing software. That I why I believe it is an "immature" technology. They are trying to fix the incompatability problems by testing through the labs but it is taking a long time.

When testing was carried out at the Citect offices in Sydney with a BACnet OPC driver, the driver had trouble communicating with this brand of controller. It was tested with two other brands with no problems. The latter two brands had no problem communicating with each other but the preferred, and eventually implemented, brand could not communicate with the other two. Curious!!!

I hope that this indicates where I am coming from. The same brand controllers had no problems with each other, of course, but when another brand or driver was introduced, there was hell to pay. Hopefully the testing labs will sort out the incompatabilities in the sam way that the ODVA have done with Device Net through the testing labs. I will not buy a Device Net device that has not been tested in this way and have never had a problem.

 

[Tripper]

If you have used BACnet in Ozz, you have probably worked for the company I speak about. They are supoosed to be the "doyens" of BACnet but had non compliant controllers and had to replace them all. The BACnet labs are testing for compliance now but I have not heard that they have progressed to a point where they are testing software. That I why I believe it is an "immature" technology. They are trying to fix the incompatability problems by testing through the labs but it is taking a long time.

I am fairly sure I know the brand you are refering to, have spent alot of time replacing it. On one particular job we were supposed to keep the services running and replace the controllers one by one without interruption. We could not keep their controllers online let alone keep services running. Bit bullet and pulled 20 sparkies in and replaced lot over 1 weekend. I think your right should be put back on to the acceditation process. Its a shame how it only takes 1 bad egg though. The computer room units were prob Atlas or Leibert.