How many Trend Logs can be achieved in any SCADA with the best possible communication Link how many (Max Trend Logs) one can have at 1 sec Update time ?
and what could be that communication Link ?
Which SCADA can do that ?
I have to log 1500 Analog values at 1 sec update time !!

Regards

 

With the best communication link, I would assume that you are using Ethernet or fibre.

Let us assuming that you will be using regular Ethernet with a transmission speed of 10 Mbps.

Also assume that you have 32 bit analogs.

1500 Analog values comprise 1500 x 32= 48000 bits (0.00048 Mbits)

This is just the value of the point. Certain protocols such as DNP implement a timestamp, and other information that describes the quality of the point. Add miscallenous overhead included to support the protocol in the native device,and the amount calculated above easily tripples.

If we assume 5*0.00048 = 0.0024 Mb to be transmitted (catering for overhead, timestamping, etc), you are still well within the bandwidth of fast Ethernet.

If my calculations and assumptions are somewhat true, you can thoeretically sample those points every 0.24 ms.

Regards,
Klaus

 

I just did 1242 analog points with a Honeywell HC900 Hybrid Controller, using the 16 Channel AI Module, (high level 4-20mA or 1-5V only) with their C50 processor. I got at least 2 updates per second.

Straight data acquisition, a front end only for hundreds of transmitters, so I only needed one AI function block per input. It is very fast to configure in engineering unis, it uses URV and LRV, not counts of the A/D.

The 1242 points needed two controllers (each controller has several racks that connect with ethernet cable). I could have stuck 960 AI on one controller (the max),

but since it needed 2 controllers, the I/O was about evenly split between the 2 controllers.

A big plus was HWIOPC, Honeywell's OPC server tweaked for the HC-900 working over the ethernet port on the controller.

Clean package, easy to configure (how fast can you click & type?), field wiring took the lions' share of install time.

Dave

 

You have to distribute in 4-5 groups with 3 ows[scada ] package having global database with Distributed database architecture.

We have done it in the past with AB PLC - PLC5/80 -2 Nos
& DH+ communication protocol & PARAGON SCADA.
No of Tags 512 AI.

Paragon Scada used to definitely give 1.0 second
trend logs for AI.

Now You can use Any ethernet based system, eg
Moore Industries , where it is possible to use
about 5-6 NCS modules & 4-5 SCADA package to
with 1.0 second trend logs.

Today's generation of client/server architecture may not meet your requirement.

Jari
[email protected]

 

Thank you all!

Dear Klaus,

You made it look very simple but whenever I spoke to any SCADA manufacturer I still have not got any confirmed reply on this.

Should we not keep in mind the time taken for the data base to commit the data on the hard disk. Would that also be a reason for the concern from the SCADA manufacturers?

Apart from data collection for logging I need to perform some control functions, and yes was thinking of DNP3 protocol as Time tagged DI information is also required for Time stamped information of various signals.

Will DNP3 act as a multi master, i.e. SCADA be one, master for DATA collection and a or many PLC CPUs be other masters for control function? CAN/DOES IT WORK THIS WAY? Which all PLCs will support DNP3 master protocol? Which all protocols (Standard known on TCP/IP flavors) send time tagged information along with value)? Can Profibius DP also send Time tagged information?
_______________________________
Dear Y KJARIWALA

Distributed Data base on SCADA I guess apart from Paragon no one gives, then how do I use any other SCADA?

Iconics I guess is another one, not so sure.

Are u telling me that we could have OPC servers on separate 2-3 PCs each having some section and then transporting this data to a single SCADA data base? Is that another way? But the final commitment on the disk level at single PC would still be an Issue that I am not so sure about.

Please help!

Thanks!

 

Here at Vista we have customers trending much larger numbers of analog points at rates of up to 1,000 per second. Of course, this requires care in
the I/O used and some design. Other customers record lower numbers, just 1,000's of analog points at rates approaching 100 per second each using Modbus/IP. All this on a single computer running Vsystem and doing other things as well. It is always nice to know that there is headroom in the system for expansion and that the products chosen will do the job!

I know that these performances are way higher than required, but one can drive a fast car slow, but one can not drive a slow car fast!

Peter

Peter Clout
Vista Control Systems, Inc.
176 Central Park Square
Los Alamos, NM 87544-4031
(505) 662-2484
FAX (505) 662-3956
Cell (505) 450-7810
[email protected]
http://www.vista-control.com

 

Would you contact me on my email id , I would
give you system architecture to achieve your
objective , unfortunately can not attach drawing or DOC file on this reply section.

But moment we use OPC , it may not be possible to give 1.0 second log .

Jari
[email protected]

 

Thanks Peter for this information,

I had a quick look at your web site, need to read more on Vista and will contact you too. Some applications done by your Integrators in steel sectors look good. Don’t know if they could share their applications and a brief write up on them.

Saw Kingfisher RTU scanner, This RTU was also in my mind. Did not see DNP3 Protocol!!

______________________________________________
| Peter Wrote:| |
|_____________| |
| but one can drive a fast car slow, |
| but one can not drive a slow car fast! |
|_____________________________________________|

But then how expensive would your fast car be!!

X-----------------------------------------------X

Dear Jari,

Will send you a mail too soon.
I have not used Paragon and will explore the possibility for using that.
Are u using native drivers for polling/scanning the I/Os?

Thanks and regards.

 

> Should we not keep in mind the time taken for the data base to commit the data on the hard disk. Would that also be a reason for the concern from the SCADA manufacturers? <

What is the typical time it takes to write data to hard disk? 14 - 30mS?

Even if we accomodate for the time required to write data to disk, there is still sufficient bandwidth in an Ethernet channel, based on the above calculation, to get 1s trend updates.

To revisit the above calculation with more realistic figures:

Channel Speed = 10 Mbps
Data = 1500 points = 1500 * 12 * 8 bits = 144000 bits (assuming 12 bytes per analog DNP point)
Assuming 20% uncertainty, 1500 analog points = 14400 * 1.2 = 172800 bits.

Propagation time for 172800 bits across 10Mbps link = 17.28ms.

> Apart from data collection for logging I need to perform some control functions, and yes was thinking of DNP3 protocol as Time tagged DI information is also required for Time stamped information of various signals. <

How fast you logic executes will dictate how often you can upate outputs. Sampling of the inputs, on the otherhand, is typically done asynchronously to the logic, but the 'new' input values will not be used in the logic until the next scan.

> Will DNP3 act as a multi master, i.e. SCADA be one, master for DATA collection and a or many PLC CPUs be other masters for control function? CAN/DOES IT WORK THIS WAY? <

Depending on the device implementing the protocol, DNP3 can act as a multi-master on an RTU. The RTU will appear as a seperate DNP address to each of the outstations.

Typically, you would have an HMI acting as the SCADA Master, with an RTU acting as a Data Concentrator, polling all the outstation on behalf of the SCADA Master. The SCADA Master will then poll the Data Concentrator for all outstation data.

>Which all PLCs will support DNP3 master protocol? Which all protocols (Standard known on TCP/IP flavors) send time tagged information along with value)? <

DNP3 an open standard so I would think that most PLCs do support the protocol. What industry are you in?

> Can Profibius DP also send Time tagged information? <

I am not very familiar with Profibus.

 

Between 300K and 1 million tags per second via our industrial historian.

[email protected]