PLC control from inlet-to-outlet

The new water treatment works have the capacity to produce 47,000m3 of drinking water per day. In addition to the standard treatment processes the plant also includes an automated processing system for removing manganese, a mineral that occurs naturally in surface water in the region.

The water treatment process consists of dissolved air flotation (DAF) clarification, granular activated carbon (GAC) filtration followed by rapid gravity filtration (RGF) and a chlorine contact tank before being pumped to a storage tank – from where it is fed into the distribution network. Although this is not the conventional process order, it is an arrangement designed to work in this area, which has the additional issue of manganese to deal with.

A crucial aspect to the control system was the use of a full dual redundant PLC system. Complete with its own CPU, the secondary system is fully wired-in and tested, increasing the system reliability by mirroring the primary CPU so that in the unlikely event of a failure of the main CPU, power supply or base unit, the secondary system can take over within 21 milliseconds from the same control point.

Andrew Robertson, technical director at Tycon Automation, the project system integrator explained that there were elements in the contract specification which meant it suited the Mitsubishi hardware. “Dee Valley Water was looking for a fully redundant system. We were able to demonstrate that we had achieved this before with a fully redundant architecture on a larger system installed on the Isle of Man, which demonstrated that could meet the performance targets and the price point set for this project.

“We spoke to Dee Valley Water at length about its preferences and reliability was of paramount concern. The Mitsubishi QnPRH PLCs are designed to work in a redundant set-up and therefore met the specification from the outset. The system was designed with two processor racks and three I/O racks in the main MCC, with one-third of the plant on each I/O rack. The system lent itself to being designed this way as the process contains three DAF lanes, six carbon filters and six rapid gravity filters which give an inherent amount of redundancy in the process; we matched with the control system design to provide the most robust engineering solution.

The pump motors are mostly controlled by variable speed drives which are connected via Profibus. A separate Profibus network was used within the MCC, with separate networks going out into the field to simplify design and increase robustness. “Mitsubishi Slice I/O with Profibus interfaces were employed to manage the I/O locally in the field, which reduces cabling and installation costs and works well with the Mitsubishi PLC that will accept most fieldbus network protocols with a plug-in comms module,” continued Robertson.

“Speed wise, and because there is a lot of digital and over 1,000 4-20mA I/O points, including flow meters, level instruments and quality instruments we segmented the Profibus networks to provide maximum redundancy and system resilience. The response time of the PLC however is far beyond what we would need for this application. PID control loops for flow control valves are well within the processing capability of the CPU, with the control loops for the chemical dosing system being the most critical. For this application we are talking seconds, rather than milliseconds, which we often work with in other more dynamic higher-speed applications.

“The Mitsubishi GX Works II software is a nice environment to work in, it is easy to find your way around and easy to use. We were able to use function blocks we had previously developed for water treatment applications which made the job simpler to realise and allowed us to add value to the project. The flexibility of the programming environment meant that the system architecture could be broken down into function blocks representing all parts of the plant. A remote access VPN link into the site means we can access the SCADA layer and therefore the PLC in real time and can easily interrogate the system to see what the raw code is doing and carry out any fault finding activities. It is important with such a large program to have a logical structure which makes it easy to interrogate.”

The QnPRH is Mitsubishi’s represents the top end of the Q Series Automation Platform, offering a high level of system redundancy to ensure complete immunity to process interruptions caused by power or system failures. This is achieved using a fully redundant architecture that duplicates processors and network links. Hot-swap capability provides an operational level of redundancy. The solution was developed specifically for use in applications where downtime cannot be tolerated for reasons of equipment damage, interruption of service, penalties, or regulatory compliance.

The dual redundant CPUs (control and standby) mean that any failure of the control CPU causes immediate transfer of control to the back-up, preventing system failure or interruption. Synchronisation of up to 100,000 words of process data is possible between CPUs per scan. Switchover time is typically around 20-40ms, ensuring a ‘bumpless’ transfer. Because the CPUs reside on physically separate racks, the control CPU can be replaced while back-up maintains system operation. Most parts are interchangeable with standard Q Series systems which controls both purchase cost and total cost of ownership.