Control.com - Nerds in Control

I have some concern of controller scan time. Scan time can be varied depend on the application, numbers of IO, data and etc. But do we actually have a standard limit for it (IEC, DIN, etc.) that are acceptably used by industries (oil & gas, manufacturing, etc.). Like SHELL, they stated that they would accept around 350ms.

There is no such thing as a standard scan time for the very good reason that there is no such thing as a standard process.

The *required* scan time of your application depends on how fast it reacts. For example, as a rule of thumb your PID loop should execute at least 2 * faster than the open loop time constant of the loop and preferably much quicker.

This means PID loops in many typical industrial processes can have a scan time of 0.5 or 1 second. However a 1 second scan time for motion control is typically far too slow - equipment moving at 2 meters per sec moves a long way in one scan.

The reality is that a blanket statement like "the scan interval should be 350ms for any and all applications" is lazy engineering and likely to get you into a lot of trouble.

Scan time is dependent upon your specific process or application requirements and the capabilities of your controller.

Rob
www[.]lymac.co.nz

It is agreed then... End-users like Shell have their own Technical Standard. I believe that there are some reasons why they state the number (350ms). They should have performed a study on their processes requirement, sensor and actuator response, safety time, SIL, etc.

Dear Vendetta,

When we talk about shell standard, it generally applies for refinery/petrochemicals. In these applications, process response is not exceptionally fast other than turbine/compressor control applications or areas where run away reactions are possible. Normally "process response time" and "process safety time" identified by licensors in PDP/BEP decides scan time requirement of control/shutdown systems in these applications. However, scan time is only a part of overall loop response time which additionally involve response time of sensors/panel components like barriers, etc./final actuation device.

For a normal petrochemical process 250 ms scan time for a shutdown system is quite acceptable.

For turbine/compressor control applications, scan time of the system are generally better than 50 msec.

Regards.

I think there is no standard for this. In my experience with one system all of the modules use the same scan rate 1 sec. And with another system all of the hardwired points are 2 sec and software points from a 3rd party system are 5 sec. Normally it depends on your CPU and your requirement. I think so.

Agree with Mr. Goswami.
However, on a side note, there is a lot of discussion by many vendors on what is a "scan" time?

My definition of a scan time is simple:

TIMEinput + TIMElogic+TIMEoutput = SCAN

Best regards,
Shahid

I have never run across any plc that had a scan time of 350 msec. The longest scan I have ever seen was just under 100 msec. You would have to really load a modern plc to get anything like that long of a scan time.

Bob,
Honeywell's HC900 hybrid controller has a fixed 500mS scan time for the analogs, something less than 30mS for DI/DO's. But they don't claim it's a PLC, either, it's a hybrid.

It works great for the thermal and level processes we run when 2 pts/sec is more than adequate.

Bud