WHY OPC F OR SYSTEMS?

OLE for Process Control – will it happen? If so, what will be the likely impact on engineering projects and systems? by Brian Tinham

February saw the launch of the OPC-Europe Council – a subsidiary of the non-profit making OPC Foundation `formed to address the unique regional and country-specific interests faced by the European Community’. So said David Rhebein of Fisher-Rosemount, its president, at the launch (C&I, March 1997, page 8). He added: `Although there will be only one global OPC specification, we want to make sure it reflects the interests and best practices from all over the world.’

Reinhold Achatz of Siemens, OPC Foundation vice president, said: `In addition, the OPC-Europe Council will manage technical support’. This is certainly germane now as the OPC technical committees gear up to move the specification from its current Release 1.0 (issued last summer) to Revision 1.01.

Due for completion by late April, this should spread the scope of OPC considerably. The committees are looking at compliance testing, historical data access, alarm handling, event logging, security, standards conventions and batch processing. They’re also liaising with ISA’s recently-formed SP95 committee, now working on standard interfaces for integrating control with enterprise systems – the middleware.

Sounds good? But what exactly is OPC? Is it going to happen? And what’s in it for us?

Well first, OPC is Microsoft’s OLE (object linking and embedding) for Process Control – harnessing Microsoft’s COM/DCOM object models and the ActiveX technologies (primarily control objects – component-ware – but also documents and scripts) to provide a single interoperability specification for all automation hardware and software world-wide. In short, sitting between data sources and data users (in the client/server model) it’s supposed to become the passport to letting all manufacturers’ offerings talk.

As for will it happen, Andrew Ballard, UK general manager of Intellution, told Sys.Build: `Yes. With the weight of companies like Fisher-Rosemount, Opto 22, Honeywell, Rockwell Software, Siemens, Toshiba and ourselves [the OPC Foundation board] behind it, plus Microsoft and about 40 OPC member companies, it’s very definitely going to happen.’

There’s no denying – it does sound impressive. And those 40 companies include big names like ABB, Iconics, Johnson-Yokogawa, Foxboro, GE Fanuc, Krohne, National Instruments, Wonderware and Schneider. Few seem willing to risk being left out.

Ballard also points to the astonishing progress of OPC compared, for example, to fieldbus and, to a lessser extent, S88 batch standards. Fact is that OPC was driven through by the original six – Fisher-Rosemount, Intellution, Intuitive Technology (now dropped), Microsoft, Opto 22 and Rockwell Software – in just 14 months. It wasn’t without some bleating from those not invited to the party; but Release 1.0 included input from over 90 other companies plus reviews by some 200 partners.

Yes, OPC is quite unlike the fieldbus debacle. Basically, the conflict of interest issues with fieldbus (be seen to be working on one – but for goodness sake don’t bring it to market!) have been preempted with OPC. Although, as with many open standards, OPC will threaten many suppliers, it’s too late. The Microsoft cat was already out of the proprietary bag – dragged by the market. We’re looking at a classic de facto.

And already the commitments are coming in. Companies like ABB, AspenTech, Digital, Eurotherm, Fisher-Rosemount, Foxboro, Gensym, Honeywell, Intellution, National Instruments, PID, Rockwell Software, Siemens, Toshiba and Wizdom Controls have all announced support – with products out now or imminent.

Intellution has launched OPC sample code onto the World Wide Web. Go to www. intellution.com.opcsample.html – downloading will let you create your own compliant server, says the company. Actually, beyond this and the SCADA Centre’s new web site at www.scada.co.uk (Sys.Build, March 1997, page 21), there’s not much on OPC – apart from the Foundation’s own site on IndustryNet.

At least there wasn’t until last month when Mike Dillamore and a few other software engineers at Eurotherm launched the OPC Programmers’ Connection – described as a resource directory (http://dspace.dial.pipex. com/opc/).

Dillamore: `We’re building OPC clients and servers and we were surprised at how little information there was, even on the Net. So we’re building the site with hot links to whatever we can find, with dialogue boxes etc.’

Clearly, OPC is becoming unstoppable. And, when you remember the might of Microsoft and its colossal vested interest in ensuring that openness goes the Windows NT, SQL Server and BackOffice way, you can be pretty certain of the future.

What’s in it for users? Openness. Ballard: `People talk about open systems, but rarely do they realise the impact of what that means.’

Essentially, it means connectivity. So long as you choose OPC-compliant kit, you’ll be able to pick any `best of breed’ device – be it server (control systems, PLCs, etc) or client (view stations, etc) – or application software, irrespective of who makes it. Also, integrating disparate existing systems should get a hell of a lot easier. Beyond these gains, you’ll also be able to harness any of the huge range of 32bit Windows applications.

As for vendors, hardware guys should only have to develop a single set of drivers to make them universally usable. And conversely, software developers should be relieved of writing drivers at all. The OPC Foundation says it’s focusing on `developing an efficient low level data access interface which is easy to understand, implement and use’.

So much for OPC. Then there’s Java – and the API for process currently being developed by Sun and a bunch of partners including ABB, Bailey, Echelon, Foxboro, Honeywell, Toshiba and Valmet. It’s not trivial to get information on what’s going on here – which is surprising since the API was supposed to be available this summer!

But according to Dan Sussman at Honeywell in Phoenix, USA: `The development is just beginning to pick up steam. Sun reports that it took longer than expected to line up internal resources necessary to get this initiative rolling in the manner that they would like. However, they are ready to drive the effort forward, and we are attempting to set up technical meetings with them.’

Talking around the industry, the following gives a balance of views. Simon Fell, senior systems engineer with Blackburn Starling, a Mitsubishi-partnered integrator in Burton on Trent: `Java’s limited in its current form. Application speed would be a problem. This might change with the JIT compilers and solid state picoJava. Also, the toolkits seem to work differently across platforms. But, the thought of being able to move from PCs up to workstations without changing code is attractive. I’m a bit sceptical at the moment, but if it turns out to be half as good as it’s cracked up to be, it’ll be good!’

Dr Bob Butts, managing director of Boward Computer Services: `Java, Java, Java; when software engineers get bored, they latch onto whatever’s new. Yes, it’s interesting, network-orientated and object-orientated, and yes it runs on any machine – it’s interpretive! The Java Virtual Machine is the interpreter. So it’s slow compared with C++. In fact, I’ve heard it described as C-! So it’s fine for monitoring and display – but control?’

Neville de Mendonca, technical director of Industrial Electronic Automation in Cardiff: `We’ve heard rumblings only. The prospect of platform independence, not just in Microsoft’s world – so that we could move up from say a Pentium to an Alpha without changing code – is appealing. And, Java appears to have a lot else going for it – like code reuse, encapsulating algorithms, no memory leakage.

`But it’s not there yet – it’s not deterministic enough. And the problem with code re-use is that it may not be entirely applicable. Anyway, few in process control want to be first; they want something that works well on someone else’s project!

`So Sun’s got a hill to climb. We may be fed up with various levels of hardware and software, but while the price-performance ratio is good, we’d have to be much more fed up to do something!’

We’ll see. As Nigel Ross, market development manager for Sun Microsystems, says: `When things change, they can change in a big way!’

So, with OPC at least for now talk of the town, I first asked Bob Butts, what’s behind all this? Ever the cynic, he said that you could see OPC as a great way of shifting the burden of driver development from SCADA suppliers to hardware vendors!

`This is by far the most expensive part of bringing general purpose, high functionality SCADA packages to market – and keeping them there’, he said. `Without that cost, companies like Intellution, US Data, Wonderware, etc will be able to make more money. They’ll say they can focus engineering development bucks exclusively on extending the MMI/SCADA/MES packages, bringing in integrated simulation, soft logic, batch tracking.

`But they’re damn well going to have to – just to hold on to the high ground. Because it’s been the need for these very mountains of drivers that’s prevented others from bringing out competing SCADA packs. Throw away that need and any old software engineer can do it.

`After all, it’s not difficult to build a SCADA package today. Put together ActiveX components with a relational database and embed them in a browser! Shrink-wrapped SCADA packages are bound to fall in price – and number. Look at the parallel in word processing. There’s only a couple of big selling packages left. And although there’s more facilities in Word than most of us are ever likely to need it only costs £150!

Maybe it’s an aside, but he’s got a point. You can’t help wondering whether certain companies that initially saw the prospect of huge growth through Microsoft compliance/alliance are now about to find the bandwagon running away from them. After all, just how many `best of breed’ systems can there be?! That’s not to say they necessarily had choice – any more than did the proprietary DCS vendors before them – driven by the then competing open PC SCADA and PLC solutions. What goes around comes around?

Asking the SCADA vendors what they thought prompted the following.

Andrew Ballard: `We have a whole team of software engineers writing drivers. And we give them away; they’re a pain. We’d much rather be developing our core products upwards downwards and across. Upwards means into the MES and ERP arena. Downwards means launching soft logic products – and there’ll be some this Spring. Across means moving, for example, SPC into SCADA.’

Was he concerned about the prospect of upstart competition? `No; definitely not. Our concern will be building and delivering best-of-breed software – recognising that users can choose, for example our OPC-compliant VisualBatch package – but someone else’s SCADA. We will be a total automation software provider.’

Adrian Wise, US Data: `We welcome OPC. Our next step is to develop an `OPC welcoming plug’, allowing the attachment of any OPC-compatible object to any of our software. It’ll be available before Summer.’

Currently, the company supports Rockwell, Siemens and Schneider PLCs with what it describes as high performance I/O drivers; there are others not in this bracket. `We’ve had to build these’, says Wise. `So have our competitors – although many have relied on DDE, netDDE, etc.

`With generic OPC-compliant drivers, driver objects will automatically be visible on-screen in our configuration environment. If you like, it means we’re enfranchising the user.’

Don Allen of Wonderware: `We’ve been portrayed as the bad guys by some – because of our open developments with netDDE. Actually, netDDE has been a success, but now we’re supporting OPC. We’re on the reviewers’ committee, and we’ll provide whatever the market wants.

`The market will vote with its dollars, and OPC is a good alternative. But I’m sure you won’t find all devices going to OPC. We’ve got about 600 drivers, some developed by our specialist distributors for quite esoteric kit. There simply won’t be the demand to re-do those for OPC.

`But SCADA applications will all become just another object in the next year or so. So we’ll be building on our two bundled all-in solutions, adding applications like simulation and asset management – reaching out to the higher levels of automation.’

Clearly, if SCADA vendors do see OPC as a serious threat to their businesses, they’re not admitting it. They’re all singing the openness mantra loud and clear!

All well and good, but do systems integrators want OPC? Will it live up to its promises? Neville de Mendonca: `OPC has huge potential. Connectivity has always caused trouble. Everything needs something bespoke to attach it – and the efficiency and quality of drivers is variable.’

Simon Fell agrees: `Although no one’s shouting for OPC yet, there should be benefits. Drivers are variable – they don’t always support what they say they do. If one supplier is responsible and we all use the OPC interface, it could solve problems.’

Mike Dillamore for the manufacturers: `It does shift the cost of building drivers back to companies like ours, but we do this anyway, particularly for less well-known SCADA packages. Also, it means that we have the control to do it properly.’

What did they say to those that claim generic OPC-based drivers won’t offer the performance of custom? De Mendonca: `Rubbish! They’re missing the point! If you’re polling an RS232 line at 9.6kB, it doesn’t matter if you’ve got a 200MHz quad Pentium Pro or a 286. Connectivity is about what information is being shared and how, and what you do with it. All we ask is that the interface behaves itself.

`The real considerations are compatibility and future-proofing. For us the price-performance ratio for Windows and Intel-based solutions is going through the roof – and we can move up this pyramid mostly without changing a line of code. Robust software you can validate, that’s what matters.

`So a standard here, with certification of compliance, would have a lot going for it. And, as OPC gains further scope and momentum, so it will go from being a useful initial building brick potentially to a cornerstone of automation.’

Will the momentum come? Fell: `It certainly could. It’s being promised by suppliers. And it should be with us in real hardware by the end of this year.’ Dillamore: `There’s a high probability that OPC will take off. We’ve had FastDDE and AdvancedDDE, but OPC looks remarkably well behaved. It will go further.’

Will OPC make connectivity too easy? Could it put systems integrators out of business? Ballard: `I wonder now about the role of integrators. As everything goes OPC-compliant many could find themselves merely in the role of consultants being paid a flat fee simply to ensure that kit specified for a project will indeed work together.’

De Mendonca: `Nothing in life is that easy, particularly where exchanging information is concerned. OPC will actually push up users’ expectations. It will make it viable for systems integrators to deliver cleverer, more flexible, more customised systems in shorter time scales – with links to multiple databases, MES servers, etc.’