Testing the water
Upgrades to the UK telephone network mean that utilities control systems will have to be upgraded for the 21st century
An initiative to upgrade the UK public telephone networks to enable the use of Internet Protocol (IP) means that a significant proportion of the communication infrastructure within the water industry, especially ones located over 5km from the exchange, may no longer work.
This move has the potential to cause major disruption as traditional business critical process control and monitoring functions are dependent upon existing PSTN (public switched telephone network) modems to provide the appropriate telemetry information.
The upgrade to PSTN in the UK was started in 2008, and is due to be completed by 2011 with similar timeframes in many other European countries.
There is a requirement within the water industry to increase the number of points within the network that are being monitored. This is being driven by ongoing compliance requirements and general expectations for an increase in proactive device management resulting in increased operational efficiency. Typically, in current systems, only outstations, water treatment works and major locations are monitored. This is set to be increased to include valves, junctions, the water pipes themselves and, in the long term, will include the meter in the home. This growth projection will increase the number of locations monitored by a typical utility from the thousands to the tens of thousands and even to the millions as advanced meters are deployed. Additional complexity is also created as a significant increase in the frequency of monitoring is also expected.
This unavoidable demand for increased monitoring means that the use of PSTN modems will not be practical, particularly for hard-to-reach locations such as the average reservoir location, which present the real requirement for underground asset monitoring.
Security by obscurity
Security within any network is mission critical. It is therefore interesting to observe that many of today's propriety process control protocols are not inherently secure and rely on a general lack of industry awareness in order to remain secure. This 'security by obscurity' approach has been successful for the water industry to date, however, it is not a sustainable technique. This is in stark contrast to modern IP-based networks where robust security protocols are enforced. Vulnerability awareness ensures we can mitigate the risk and protect against any instances where security could be compromised.
There is now a strong move towards using open standards for the communication mechanisms within the process control network. This is typically based on IP, for example DNP3 over TCP/IP is now widely available. Many process control equipment manufacturers are embedding both Ethernet and IP stacks into their offerings allowing easy integration into existing data networks. This allows companies to make use of the cost-effective and widely available broadband infrastructure across the UK and much of Europe.
Increasing the frequency and number of monitoring points in the water network leveraging always-on connectivity makes real-time data gathering across the water network a reality. This will enable allow water companies to:
- Proactively manage faults;
- Significantly increase the ability to provide proactive leak detection;
- Ensure greater customer satisfaction;
- Reduce water wastage;
- Accurately model the water network;
- Predict demand trends and issues;
- Utilise process automation techniques.
Modelling will bring additional benefits, such as the ability to ensure pumps and motors are operating as efficiently as possible providing substantial sustainability benefits from reduced energy usage and reduced green house gas emissions. The use of appropriate pump scheduling has been shown to give typical energy cost savings of 10-15 per cent, which is significant when considering that 95 per cent of the life cycle cost of a pump is the electricity used and only 2 per cent is the cost of the hardware.
New methods for process control are starting to appear in the market, requiring additional network capabilities that are not provided by traditional telemetry networks using PSTN modems. Video technologies and analytics can be used over an IP network to analyse water quality or flow rates in remote locations where probes are not feasible. Physical site monitor--ing, site security and CCTV can be integrated into the corporate security systems rather than today's standalone systems.
21st century options
The trends and issues outlined highlight that many of the current Water Utility networks need to be upgraded in some manner in the next asset management plan period. There four main options available.
1 Tactical Legacy Modem Upgrade (limited lifespan) This would entail replacing existing outdated modem infrastructure with new modems that are tested and certified with the next generation IP enabled PSTN voice infrastructure.
It may also be necessary to replace some of these modems with cellular technology such as GSM (Global System for Mobile) for remote locations that may not be able to be supported using traditional PSTN modems. The use of newer GPRS (General Packet Radio Service) or HSDPA (High-Speed Download Packet access) services would not be possible unless newer IP enabled process control and monitoring equipment was available. Just replacing modems with updated models would be a tactical approach and will not enable many of the new process control techniques that are being demanded by the water industry and would not enable some of the new business functions and ways of working that we will talk about in a subsequent article. This is not to say that modem technologies such as HSDPA, GPRS and even PSTN will not play a part in the 21st century water utility network, but simply replacing old legacy modems in a like for like basis would be of limited benefit and have a limited lifespan.
2 Tunnel telemetry traffic over IP (migration step) It would be impractical to remove all existing serial-based telemetry protocols overnight and replace them with an IP infrastructure, however, as already discussed using IP as the backbone can significantly enhance the process control network. Existing telemetry traffic can be tunnelled over the IP network using techniques such as reverse telnet, dedicated circuit emulation hardware or specific tunnelling protocols such as UDP Telnet (UDPTN).
UDPTN is a feature that allows asynchronous serial traffic to be encapsulated into User Datagram Protocol (UDP) packets and then send this data without needing to establish a connection with a receiving device.
The asynchronous serial traffic over UDP feature provides a low-bandwidth, low-maintenance method to deliver data.
These techniques will allow the migration of existing services onto the IP infrastructure without the need to rewrite the applications or change the end systems. While this technique could be a more permanent solution than the first option, it would ultimately be used as a stepping stone to a fully converged process control network. Tunnelling the traffic would allow sites to be migrated in a controlled manner away from their existing serial based telemetry protocols onto the IP converged network. Tunnelling the traffic may still have some of the capacity issues we have today, we still need to manage multiple serial lines as we are not converting the traffic but simply tunnelling it. Port contention may be seen as an issue, especially in an always-on monitoring solution, and as such we may not realise all the benefits that a fully converged network would bring.
3 Convert telemetry traffic to IP routable traffic (partially converged) As stated in the previous option it would be impractical to remove all existing serial based telemetry protocols over night. Rather than tunnelling the traffic over the IP network we can convert it from a serial protocol into a standard based IP protocol that can be routed over the IP data network.
There are over 300 protocol translations for SCADA (Supervisory Control and Data Acquisition), telemetry and process control traffic. Once the serial data is converted into IP it can either be left in this format for IP enabled front-end servers or converted back into serial data to allow easy migrations.
The conversion process can be undertaken within the IP router as well as using external protocol translation appliances. An additional benefit of translating the traffic means we terminate the serial protocol on the router or appliance and as such we are less concerned with how the serial protocol will perform over the network.
This also removes many of the scalability issues associated with serial lines we see in the first two options. We now have an obvious migration path directly into a fully converged process control infrastructure running on an IP backbone.
4 End-to-End IP (fully converged) Within the fully converged network all end points such as PLCs (Programmable Logic Controller) or outstations would be IP enabled and no protocol translation would be required. During the migration phase to this infrastructure it is likely that traffic may be tunnelled or converted until all the end points are IP enabled, we will therefore see a mixture of option 2/3 and 4 during a transition phase. The stages we are seeing in the evolution of telemetry networks are very similar to the evolution of mainframe communication that started with serial connections, then moved to using DLSW (Data-Link Switching) to tunnel the traffic over IP and finally to IP enabled mainframes and terminals. We now take IP as the mainframe connectivity protocol as a mater of course.
We need to ensure that we create a single converged physical network that can be logically segmented for various types of traffic or groups of users. Typically, data would be treated in a different manner to that of voice traffic, process control traffic and physical surveillance traffic. While all of these traffic types would be running over the same physical infrastructure, they would have different quality of service and security mechanisms implemented. Voice traffic is very loss- and delay- sensitive but requires a relatively small amount of bandwidth; this is similar to process control traffic which today is low bandwidth and loss sensitive.
The move to IP will ensure the protection of the process control traffic from a security perspective. This logical network would be segmented with firewalls from the other networks. We may choose to implement robust security protocols as well as putting mechanisms in place to detect potential attacks.
The benefits of a converged IP infrastructure become obvious when considering the number of different devices in a treatment works. Back office functionality, corporate access, process control, process monitoring, telephony, wireless telephony, security alarms and CCTV all ckamour for attention. Each one has traditionally had a dedicated connection back to the data centre. These connections would use multiple communication mechanisms, such as PSTN, corporate WAN, GPRS and satellite, and in some circumstances there may be multiple instances of each.
Bringing all these back over the IP network does not kick these diverse communications mechanisms into touch; it does however mean they are now controlled from one place - the router. This router makes the decision about the best communications medium to use based on traffic type, bandwidth, time of day, network availability, and so on. We thus make more efficient use of our converged network and also normalise the in-building communication infrastructure making new service deployment faster and much easier.
It is legitimate to compare the water utility company to a very large manufacturing plant with two main processes - the process of making drinking water and the process of cleaning the wastewater. The monitoring and control requirements are similar to those of any manufacturing plant, the main difference coming from the geographical area of this water manufacturing plant.
As in any manufacturing business, it is important to monitor many of the points in the manufacturing and supply process. Today the water industry only monitors major processes and none of the intermediate points, this is especially true of the water distribution network. The converged infrastructure will enable the high level of monitoring required and will also bring many other business benefits. We will look at these in the next issue of E&T.