WBF Barcelona

I case you are wondering, I shall indeed be going to this event, I am looking forward to it.

See you there I hope.

Alarms and Equipment States

How many reports on hazardous incidents have you read about where the alarms presented to the operators were excessive and the resulting confusion contributed to the incident.

I was reminded by this article

Why Is Safety so HARD?

The problem has been known and understood for decades, and now we all know that alarms should be suppressed when they are not relevant, and that when they are they should be prioritised.

But engineering a solution is time consuming and expensive.

The solution has to be specified, reviewed and approved, and maintained as operating experience is gained and the solution is modified, via an approved change process of course.

Typically these are described by text, cause and effect matrices and logic diagrams.  

Mostly the cause and effect matrices describe the responses to potential hazardous events, such as process upsets. The matrices do not often cover the alarms, although some do mention them typically in notes. Some control system actually support Cause and Effect matrix based design, and can translate them into control logic. For example Siemens has one

These are much more constrained than the typical excel version that people produce.

 

Using a state model provides a highly efficient way to define the enabling of alarms. The safety system, as a complete entity is defined in terms of possible states, a method that vastly reduces the number of states that have to be considered.

Then each possible alarm can be considered for it relevance in each state, producing an Alarm State Matrix. 

ISA–TR88.00.02

Dave Chappell's blog reports that "much anticipated Technical report from Make2Pack"  is now available, on the Part 5 blog and indeed subscribers were sent a copy, which I have been perusing. 

It is a good document, but not a lot to do with S88. 

I did much more than lurk on the postings, I spent many hours reading drafts and making comments about it, to try to align it with the real Part 1, and I took to using the phrase ~Save the Batch during the debate. It got quite intense, and at one point I was told "Better keep looking over your shoulder …the S88 police are coming!" 

Needless to say that did not deflect me. 

I think that I changed it in several respects - challenge is essential.

I objected initially to calling a machine a Unit as in most respects machines are little more than control modules, albeit complex ones (but only internally, not for the recipe). 

I accepted that they could be called Units because each machine in a filling line might be processing a different batch, thanks Dennis for reminding me of that and the parallel with continuous processes.

I did not accept that terms such a unit procedure or phases should be used in the report, as they were originally, and they are not there now.

S88 Graphics

I have seen many automation system HMI's. Often the process graphics have been produced based on the P&ID's. But P&ID's are not normally a good basis for graphics as they are not designed for operational purposes. Using the S88 Hierarchy can be much better.
There is at least one plant where the S88 Physical Hierarchy is used as the basic for the HMI graphics structure.
I believe this is a good structure, not only because it makes for simple uncluttered graphics but also because it exposes the S88 modules of a system to the operators.
Here is an example
The Plant overview shows a graphic that contains several Process Cells
Clicking on a Process Cell object opens the Process Cell diagram, which shows all the Units in the cell. Note - these can use colour and text to indicate the status of each Unit.
Clicking on a Unit object opens a Unit diagram
Note that this does not show all the valves etc in the Unit, it just shows simple objects for each equipment module but again with colour and text to highlight status of the equipment modules

Clicking on an equipment module object then opens an equipment module diagram where you can see the valves etc in the em.

From here you can also go to an Equipment Module Faceplate

Or of course a to Control Module Faceplate.

In this case, the automation for the plant was designed with ControlDraw, and the graphics correspond to relevant objects (diagrams) in the ControlDraw model.

Comments are as ever welcome.

Download S88 Part 1

Yes, you can download the entire ISA-88 (S88) Part 1 standard for free, the original and best version.

- note - this has now disappeared 
I mention this because it is one of the most common searches on the controldraw web. Of course it is not there, it is not supposed to be free - though I think it should be.
I did not put it there, but someone (not me) has put a link on the Wikipedia S88 section. I found it just 2 clicks from Jim Cahill's blog - which is worth reading by the way.
So get it while you can.
Updated May 2011
Some Chinese person has now posted the standard here
http://wendang.baidu.com/view/148a9bb91a37f111f1855be3.html

More on ISA S.5.06 Functional Requirements

I have spent some time delving into this standard, and the examples it provides.
Further than that, I have now created a ControlDraw model that captures basically everything that the standard shows. This is for me the best way of understanding a process.


One thing that amused me about the batch reactor example is that it has nowhere to send the batch it makes apart from down the drain! It calls dumping the product down the drain the Dump Phase, part of the Transfer operation, I suppose dump is appropriate, but Transfer? Maybe I missed something - here is the P&ID




S5.6 has this thing called a Sequence matrix. This is a condensed overview of the example process, but I suspect that it really would not be workable when scaled up to a real, and more complex process. Furthermore it conflates too much into too little, and if you are trying to do in in Excel demands lots of formatting. ( And the example manages to avoid the most complex issue, the transfer.)

So my model does not use that method - it is optional, you can also use Sequence charts - and instead uses those, in the form of SFC's and PFC's
Maybe a complex query could generate the matrix from the combination of the sequences and the equipment - it is possible but it would be a huge table compared to the example sequence matrix. Even without adding a real transfer!

Will S88 improve - or should it?

Will ongoing developments of S88, specifically Parts 1 and 5, improve automation?
I am far from sure, in fact I think they will not.
I think it is arguable that even the original Part 1, for all the applause it has received, has actually had the effect of damaging the development of innovative automation products.
Why do I say that?
For a start, the ‘Procedural State Transition’ model given in the original Part 1 (which does not need updating imho) as an example is but one of many possible alternatives, and certainly not the ‘best’. And it may well have constrained product evolution. There are plenty of other similar issues, not least the handling of common resources that the same could apply to.
And now, in the Part 1 update and Part 5 the standards committee is further limiting opportunities for designers to come up with something much better, by refining that which did not need refining.
Standards have their place, we would hardly be able to function without standards such as the metric one for measuring, but S88 is not in that domain at all, there is no science behind it yet, and no sign of one emerging.