Make sense of data to boost profit01 December 2005

Significant advances are constantly being made in gathering information about what is really going on in process and industrial plant - especially where located in more remote locations - and in processing this information to reduce downtime and improve profitability.

Much of the leading-edge development in sensing is to be found in small companies, often spun out of university research teams.

Dr Jack Hale, based in the School of Mechanical and Systems Engineering at the University of Newcastle, and his former associate, Dr Robin Stephenson, have set up a company, Hale-Stephenson, to commercialise their expertise in monitoring remote plant, Specifically, the purpose is to pinpoint those rare events that operators need to be advised of urgently, so they are not deluged with information they do not need to look at.

At a recent conference, 'Sensors and their applications XIII', organised by the Institute of Physics and held at Greenwich University in Chatham, Dr Hale revealed the type of events he has in mind: ships colliding with wind turbines, land slips affecting pipelines, significant oil leaks and acts of terrorism.

When something does occur, the operators need to be alerted quickly and must be able to understand what has happened. This means gathering information at a high rate, so transient events are logged in detail, including what led up to them occurring, although only events of significance are recorded and forwarded.

The system comprises several layers. A Local Monitoring Computer (LMC) is sited at the wind turbine, pipeline or other remote plant to be monitored. The LMC is a self-contained, battery-powered unit that can monitor eight channels, which can be sampled at any rate up to 2,000 per second. Any sensor type may be used, but for structural monitoring might include strain gauges, geophones, vibrometers and accelerometers.

The LMC reports to a Remote Supervisor Computer (RSC), via a GSM mobile phone link. The LMC monitors the sensors continuously, using very little power. The GSM module, on the other hand, has a much greater power requirement and would quickly drain even a large battery, if used continuously. So the solution is to establish the wireless link only when it is required - ie, when the LMC detects an event greater than a predetermined threshold and needs to transmit this fact, along with the monitoring data immediately before it. The GSM module is switched off at all other times. In case the LMC fails without anyone realising, it also performs a self-diagnostic health check at regular intervals and signals this fact to the RSC. Should the health check signal not be received at the appropriate time, the RSC automatically flags a warning.

The RSC can accommodate any number of LMCs. It monitors the GSM continuously, waiting for a contact from an LMC and, when contact is made, downloads a data file, which is then linked to a special internet page for transfer to the Client Service Computer (CSC) for analysis, event logging, archiving and alarm generation. While the link is active, the RSC also sends the LMC any changes to operating parameters, such as threshold levels and sampling rate. System control is exercised by an administrator, who has access to the RSC and the LMCs. Clients do not normally have this level of access, which is potentially damaging to the system, but have full access to any data generated by their own LMCs.

The LMC, and its associated software, was developed by Design Unit, a design and consultancy organisation within Newcastle University. In addition to a computer optimised for high-speed data logging, it also contains a GPS satellite positioning module and a short-range radio link for future mobile and multi-LMC applications. It is built into a robust, watertight extruded aluminium enclosure approximately 100x70x30mm. Hale-Stephenson acknowledges that the development of its technology was only made possible thanks to receiving a DTI SMART Award. Capital is now being raised to take the development to its next phase.

Novel sensors to feed data into such systems were the main subject of the conference. Dr David Lamb of the University of New England in Australia discussed how to improve the control of red wine production by devising a method of measuring the red colour in the presence of large amounts of suspended matter, as might be caused by grape skins.

The method developed, in conjunction with Dr Yeremias Bunganaen (since returned to Indonesia) and Dr Peter Lye, is to use a piece of optical fibre, with part of the cladding removed. This permits the evanescent field from the light going down the fibre to interact with the substances in which the fibre is removed. Interaction with the chromophores, the red colour in the case of wine, results in the attenuation of the intensity of the light travelling down the fibre core. The limited penetration depth of the evanescent field - about 1 micron - is insensitive to the scattering by suspended particles, provided they are larger in size than the field depth penetration.

Another idea that appealed to even more of those present, mainly because it was potentially so cheap, was to use inkjet-printed 'fuses' that responded to the presence of other chemicals. The idea was presented by Dr Muhammad Mabrook of the School of Engineering and Centre for Molecular and Nanoscale Electronics at Durham University.

The work seemed mainly to focus on the inkjet printing of PEDOT-PSS, an electrically conducting polymer, and subsequent measurement of the changes in the conductivity of printed layers in the presence of alcohol. Thick films made by printing four or five layers on top of each other increased their electrical conductivities in the presence of alcohol, while thin films, made by printing one or two layers, dropped abruptly.

The mechanism is thought to be based on the physical behaviour of the printed chemical droplets making up the fuse. If there are lots of droplets, the addition of alcohol tends to spread the film, improving conductivity. If there are relatively few droplets, they tend to ball up and separate.

The intended applications are for breathalysers and 'Alcolocks' for cars, but in fact the basic idea could be used to detect any chemical vapour that could be made to interact with an inkjet-printed organic thin film.
One message that came from the conference was that it was possible to come with a clever little device that could detect and/or measure virtually anything. If any of our readers need assistance with plant control, we suggest that they contact the Instrument Science and Technology Group of the Institute of Physics, which organised the conference. If the Group doesn't know somebody who already has the required technology, they will almost certainly know of somebody in the UK who can invent it!

The profit factor

However, having discovered the means to measure exactly what is happening in a plant or process, the challenge then is to use the gathered information to improve profitability.

This entails processing the information in real time, with the goals of reducing downtime, running costs, operating costs, waste and rejects, re-work, and maintenance costs. This usually requires detecting problems before they become serious, and recording stoppages and dips in performance. Once problems have been located, they can be investigated and linked to other occurrences to ascertain their causes. In a fairly simple operation, this could be achieved by keeping records of problems in a spreadsheet, which can then either be referred to or turned into charts to aid analysis.

For large or complex operations, it may be more beneficial to use specialist software. One particular product that is aimed at addressing such tasks is Ampla, from Citect, an Australian company headquartered in Sydney, which seems to have achieved some spectacular successes in the mining industries and is endeavouring to expand its activities into other sectors.

Ampla is described as an ISA-95 compliant scaleable suite of Manufacturing Execution Systems (MES) solutions. The ISA S95 standard for enterprise-control system integration defines interfaces between applications at the Industrial Control Level and applications at the MES level. As such, it bridges the gap between control level communications, which are typically implemented using OPC, and business-level applications using EDI and various B2B standards. The standard is being defined in 'parts', which are being individually released as they are approved.

Ampla is the latest embodiment of the MES software offered by Citect. Its name comes from the Latin verb, amplio, to improve. Its aim is to gather data across all levels of an enterprise, perform data analysis and present the results in what are easily understandable formats.

As Jeremy Bolton, the company's sales and marketing manager, says, "this is not a cottage industry product". The launch pack is for a minimum of five users, with costs starting at £10,000 and running up into the £millions. Citect customers include Nissan in Sunderland, but the company's star customer is apparently Implats - Impala Platinum in South Africa - which has 200 users and is said to have saved £250,000 per day since implementing this software. Implats currently produces around 1.7 million ounces of platinum, which equates to 25% of global production.

Ampla sits on top of an Oracle or SQL database and lies between the Supervisory Control And Data Acquisition (SCADA) system, which gathers information from sensors, and the Enterprise Resource Planning (ERP) system. One of the main aims is to improve Operational Equipment Effectiveness (OEE), defined as availability x performance rate x quality rate. Availability is running time divided by scheduled time. Performance rate is actual output divided by target output. Quality rate is good output divided by target output.

While the three individual quantities are typically each around 90%, multiplying them together produces a figure of just over 70% - indeed, OEE is often 60% or worse. Since this represents the percentage of what the plant is producing, as opposed to what it could be producing, relatively modest improvements in performance can result in quite significant improvements in profits.

Ampla comes in nine modules: Production, Quality, Tracking, Downtime, Cost, Maintenance, Knowledge, Planner and Metrics. However, OEE improvement can be achieved by purchasing only three: Production, Downtime and Metrics, in a package named Ampla Performance.


Key to best practice

When monitoring remote plant, rare events that operators need to be advised of urgently are pinpointed.

Transient events are logged in detail, although only events of significance are recorded and forwarded.

Once problems have been located, they can be investigated and linked to other occurrences to ascertain their causes. An idea that appealed to those present was to use inkjet-printed 'fuses' that responded to the presence of other chemicals.

SOE

This material is protected by MA Business copyright
See Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the sales team.