To be frank, I am not qualified to answer this question. In my uninformed opinion, however, I would imagine that penetration here would be low due to the typical real-time needs of these applications.
The distributed nature of typical industrial control systems isn't in itself a problem for Web services. In fact, Web services are well-suited to highly-distributed, heterogenous applications. However, a significant problem may be the real-time nature of many industrial control systems.
Web services are very good at integrating diverse IT systems and applications together: XML, SOAP and WSDL provide a standards-based approach to connecting almost any application, no matter what language it was written in, what platform it runs on, or the underlying transport protocol. They also provide additional flexibility and cost advantages over traditional integration approaches. Web services an also offer secure connections.
In embedded control applications or for many complex multivariate systems, often all that is required is remote monitoring, control and data logging, and in these situations Web services would work very well. Web services are also excellent for knitting together existing controls systems - for example, two or three existing plants or processes that need to be centrally managed or co-ordinated.
When designing real-time or near real-time applications you need to worry about synchonization, latency, jitter, and topology restrictions. SOAP-based communications are inherently verbose, and in these situations Web services may not be the best approach. This has probably impeded adoption in the industry.
My advice would be to look at the application requirements closely and select a simple pilot program. Start small, with a single point contol system. Web services are so cost-effective and flexible, you will find numerous applications where they will work very well.