Wiley Wastewater Quality Monitoring and Treatment_20 pot

Pre-normative work Standardised testing of on-line sensors/analysers, Anders Lynggaard-Jensen, ISO Working Group under an EU project in the late 1990s was reported at the ISA-Tech Confer

Data Quality and Comparability 357

rconductivity;

rdissolved oxygen;

rturbidity;

rammonia;

rnitrate;

rnitrite;

rorthophosphate;

rchlorophyll a.

A consultation document was released at the end of 2005 for the following portable water quality measurements:

rtemperature;

rpH;

rconductivity;

rdissolved oxygen;

rturbidity;

rammonia;

rnitrate;

rnitrite;

rorthophosphate;

rChlorophyll a.

This particular consultation document comes from the consolidation of extensive work on earlier work for a workable format for ensuring high quality chemical mea-surements individual measurands It is thus likely to be the basis of an Environment Agency mandatory approach to administering their self monitoring policy

MCERTs is only an input for part of the uncertainty chain IPPC (Integrated Pollution Prevention and Control) and PPC (Pollution Prevention and Control) do not now rely on ‘end of pipe’ measurements and it is to be expected that audit trails and uncertainty chains will increasingly be demanded for key process variables Calibration, maintenance, training and consumables management practices and controls form an integral part of understanding the quality of measurement data Obviously there is a lot of judgement and reliance on skills involved and it becomes clear that comparability of data is a challenge

Pre-normative work (Standardised testing of on-line sensors/analysers, Anders Lynggaard-Jensen, ISO Working Group) under an EU project in the late 1990s was reported at the ISA-Tech Conference at Interkama in October 1999:

In order to support the protection of the environment there is an urgent need for im-provement of comparability, reliability and quality of measurements obtained from in-situ on-line sensors/analysers used to determine the composition and flow of effluents from wastewater treatment plants and industries Users cannot check and compare different products before purchase without individually completing costly experimental trials and

so the number of potential users is restricted Furthermore, there are no standardised procedures to provide users and regulators with acceptable and robust validation of compliance data.

As a result, the widespread adoption of systems is limited and the benefits of this in terms of more efficient and effective pollution control are lost Moreover, the instrumen-tation industry does not have any standard to develop systems to, and so on the one hand

is unable to properly define its development costs and on the other hand has no means

of independently demonstrating that its products are ‘fit for purpose’ in the European market.

The paper described a pre-normative research work and the subsequent work in a working group under ISO concerning standardisation of in-situ on-line measurements

of water quality determinands The work includes development of a draft test proto-col for validation of the performance of in-situ on-line sensors/analysers to guarantee comparability of results, the practical testing of the draft test protocol to access its appli-cability to provide a final test protocol and turning the final test protocol into an agreed standard/guideline.

The EU thus recognised that there was no comparability of data between Member States A current EU 6th Framework Project to deal with this problem is SWIFT (Screening Methods for Water Data Information in Support of the Implementation

of the Water Framework Directive; www.swift-wfd.com)

Establishing cost effective and useful measurements requires significant inputs from a wide range of disciplines:

restablish application need;

rdesign application and data flow;

rprocurement;

rinstallation;

rcalibration;

rmaintenance;

raudit trail.

An UKWIR Report in 2000, ‘On-line instrumentation Standards and Practices Ref

No 00/PC/03/1’ concluded, among other things, that measurement was below se-nior management’s horizons in the user industry and recommended the generation

Disparate and Sparse Data 359

of multidepartment understanding and cooperation Dealing with widespread sparse data will be even more of a challenge and will require significant changes in man-agement structures skill sets and cooperation in the users

During the UKWIR project it was discovered that at least one UK water company switched off two archives because there was no reasonable way to normalise the data with other data and the future format

The quality control of the transmission and archiving of measurement data is a further part of data management if comparability of data is to be achieved

6.1.6 DISPARATE AND SPARSE DATA

Up to this point the generation of data has been from man-made devices (test kits, portable instruments, samples to laboratories, etc.) These data generators may be widely spread (sparse) in, for example, a sewer network or water distribution scheme

or for the environmental control of a river catchment As we have already seen the methods of generating data are many and various and time lines may be difficult

to establish Data derived from test kits may be occasional (sparse) and since the operators may vary or change substantially over time it may be difficult to estimate uncertainties for that data

However, for large processes, particularly environmental, there are other inputs that will affect the merging of data Diurnal changes and weather conditions are known to affect the generation of data but are frequently ignored Other disturbing events may occur and may not have been recorded

SCADA data are part of the information being analysed as part of a EU ‘Mi-croRisk’ project whose primary aim is to undertake quantitative microbial risk as-sessment on 12 complete water supply systems (source/catchment, treatment, distri-bution, consumption, infection) – essentially Monte Carlo style simulations of the doses of pathogens that consumers receive in model full scale systems Obtaining high quality data is turning out to be much more of a challenge than was expected

A considerable amount of such data and analogous high resolution data sets (e.g microbiological data) collected daily for several years is available However, sys-tematic inspection of SCADA data and associated diary data from a particular water treatment plant has proved much slower and more problematic than expected Several issues have arisen

(1) SCADA data management with high resolution data sets quickly exceeds the capacity of MS Access to manage and distilling the raw data takes a long time Also, there do not seem to be many standard/statistical algorithms for distin-guishing baseline from event data in long time series It seems that eyeballing is the most effective technique

(2) SCADA data are often provided as stand alone without clear links to what is actually happening in a water treatment plant

(3) It is often not clear whether departures from targets represent problems or main-tenance or poor instruments

(4) Measurement facilities are often not well maintained and/or the quality control regime for them is not well documented

The performance of water treatment plants is often not recorded electronically and sifting through paper records makes the task of quantifying and understanding hazardous events very time consuming and difficult The use of SCADA data is obvious but it is important to recognise its limitations in the role of risk assessment After all the data are mainly for real time process control therefore their interpretation for hazard analysis requires care Although not yet established with the regulators and courts, there is a growing awareness that on-line data can be legally robust Samples analysed in a laboratory are the established source of regulatory or other control data; despite the known potential problems with the relevance of the sample taken, transport to the point of analysis and the quality of the determination itself Establishing a sound audit trail is still a challenge and, surprisingly, it is not usual for a measurement uncertainty chain, based on the full sample to data presentation trail, to be available For many chemical measurements Blue Book methods are not widely accepted, not available, or more than one method is available

Underlying this all is the ‘know your system’ philosophy being promoted in the new ‘Water Safety Plan’ approach for managing water supply risks

The above discussion has concentrated on a process for which it is reasonable

to expect a reasonable amount of data; although some of it may be from disparate sources In this EU project the instrument data and the diaries need to be matched

If we broaden the information needed to company wide or environmental con-siderations it can rapidly be seen that for economic reasons the data that must be merged may be sparse and disparate

Ways will need to be developed to instil confidence for infrequent users of the data such as senior management, pressure groups and regulators The data need to

be presented with quality statements, which is a challenge, but there is also a urgent need to find graphical methods to provide confidence that the complete ‘system’ is

in balance Senior management, pressure groups and regulators do not have the time, and possibly the expertise, to deal with all the data Concepts like ‘mass balance’ and ‘ecological stability’ can be used to provide a framework against which the detail might be considered Most factory HMIs (human machine interfaces) allow

a selection of a number of measurements to be overlaid (Figure 6.1.2) Some of the measurements are continuous and some are infrequent samples analysed in the laboratory The traces would be selected so that if the particular combination is

‘stable’ then it is a reasonable assumption that the other measurements are within acceptable bounds It is a relatively simple matter to train a wide range of people to understand what is normal and what needs closer inspection The UK Environment Agency has already declared a policy of self monitoring so it is to the advantage of

‘polluters’ to demonstrate and communicate indicative information with the detail

Research and Development Support 361

Particulates (Dust)

Oxides of Nitrogen (NOx)

Sulfur Dioxide SLF used at 4 t.p.h SLF not used

2000

1500

1000

500

10 20 30 40 50

Figure 6.1.2 Tracking operational performance with overlaid measurement traces

readily available if necessary The timescale can be rapidly changed so that both incidents and long term trends can be inspected or studied

The WFD implies a massive need for data, which is unlikely to be met Many believe that modelling is the only viable approach and the selection and quality of indicative data will be the pacing issue It will be challenging to ensure that the users

of the resulting information have a comfortable feel for the quality of the underlying data and some idea where improvements in measurement and in modelling might be made and to what effect

Elsewhere it is argued that information will increasingly come from the merging

of widespread and disparate data in the form of models In this case measurements may be calculated and inferred and it is probable that some of the input data may have a high uncertainty but adequate for the needs; with consequent cost savings

6.1.7 RESEARCH AND DEVELOPMENT SUPPORT

The very large instrumentation companies have provided support for long-term en-abling technologies; some of which only became ‘open’ after the market had been es-tablished, i.e HART protocol Some long-term support, such as the Foxboro/Invensys work towards ‘self diagnostic’ sensors at Oxford University has yet to gain com-mercial rewards but spin-off work on the way seems to have justified the effort A common framework for self diagnostic sensors might well be an important donation

to the worldwide sensors industry

It is not unknown for a major user to spend a considerable amount of money over many years and apparently have nothing material to show for it Often this happens because of people mobility but more often because of unclear objectives However, closer inspection often uncovers benefits that are real, but hard to quantify; ideas and prejudices challenged, broader perspectives considered, etc

The interoperability of telemetry systems is being taken seriously by the sensor and telemetry community and a cross industry activity (WITS; www.ukwits.org) is already making progress on common methodology A methodology for communica-tion from the management network out to the sites will be undergoing interoperability trials in 2006

A cooperation exercise called WASP (www.wasp-protocol.com) is starting to make headway at the next level down so as to bring some sort of order to the con-nection of the WITS level to the sensors and actuators Generally the WITS level is driven by information technology people whereas the WASP level is driven by in-strumentation people and the two cultures need to develop a common understanding

of the issues

Of course all is not sweetness and light, as the decade or more of inter company and country rivalry in the Fieldbus wars showed However, some enlightenment is showing and real progress is being made on the various modes of serial communi-cation The breakthrough was probably the broader acceptance that a single serial communication (fieldbus) was never going to happen and inter-operability was a better objective The uptake of Ethernet and the internet was the final nail in the coffin of formal international standards having any precedence Standards setting with timescales of a decade or more cannot be relevant when viewed against the very high rate of change of serial communication technologies

6.1.8 USERS

The users, whether industrial concerns, regulators or governments, are usually com-partmentalised by discipline or market sector, whereas measurement is usually a multidepartment, multidiscipline activity The users have little or no wish to become engaged with the sensor technologies The days of the separately identifiable in-strument department or senior engineer are long gone in the majority of industries There are often pockets of identifiable expertise but these will disappear with time since industrial on-line measurement is rarely an identifiable higher education topic On-line measurement is much less than 1 % of capital and does not figure in senior management planning unless a commercially viable project plan is placed before the project selection committee

Measurement decisions may well end up at the end of a long line of subcontractors

as part of the outsourcing approach to business This generates problems for the in-strument supplier since the decision making unit (DMU in marketing terminology) is often large and diffuse with quite a lot of mobility in its constituent parts Who do you approach and convince with a new idea and who has the budget to evaluate the idea?

Users 363

It is easy to see that there is a ‘disconnect’ between the suppliers and the users Each has an honest and hard working approach to the needs and responsibilities but the business and cultural differences need to be addressed

A further problem is that although there is a now grudging acceptance of the need for ‘maintenance’ and other OPEX issues the out-sourcing and multidepartment nature of the user makes budgeting, coordination and resource allocation close to impossible

A normal asset life is taken as 10 years or more and this is challenging for the fast moving technologies that underpin industrial on-line sensors For example electronic chip, embedded computers and serial communications are all moving with much shorter time constants than 10 years Given that an industrial on-line sensor may take several years from inception to established commercial viability the planners, developers and marketing departments have an interesting challenge

An idea mooted some years ago and progressed by some suppliers was to lease the

‘measurement’ The supplier takes total responsibility for the supply, installation, calibration, maintenance, updating, etc This approach has been used for decades in the North Sea oil and gas production industry

It is increasingly difficult for automation suppliers to sell ‘boxes’ The majority

of suppliers have recognised that they have to sell ‘solutions’ However for small value items it is difficult to achieve the critical mass where the overheads become

a small part of the total cost Of course the user still has to be satisfied that the measurement(s) has a business benefit and has to take ultimate responsibility for the data, which is a nontrivial challenge

A simple solution for individual measuring instruments is for the supplier to design for ‘fit and forget’ and ‘point and shoot’ In other industries, such as medical, auto and home goods, there seems to be no alternative and this may be the only approach for demands such as monitoring sewers for the onset of flooding or the monitoring of drinking water supply networks for unwanted substances or the monitoring of river catchments (Figure 6.1.3)

Modems situated fairly close to the sensors Possible peer to peer cooperation

Radio Battery Secondary electronics Sensor

Figure 6.1.3 Structure of an autonomous instrument

Available technology already allows autonomous on-line instruments, as in Fig-ure 6.1.3, with a life of 5–10 years and some are being trialled We can expect that the Homeland Security and Battle Field industries will provide the fundamental research and useful components The problem will be gaining adoption in process industries with long asset lives and well established methodologies and cultures

In the past sensors and measurements have been specified without proper cost– benefit analysis with a high proportion of instruments being improperly installed and supported with proper OPEX budgets and defined ownership A 2000 report for the water industry (UWKWIR Ref 00/PC/03/1) estimated that at much as 25 % of the installed instrument base could be removed with consequent cost advantages It was suggested that a further advantage would be an increased confidence in the remaining measurements Action was taken and there is a notable increase in attention for the measurements that directly affect performance

It is generally accepted that all environmental measurements are driven by regu-lations or other socially driven requirements The user is understandably reluctant to make measurements that might then be used against it or to make improvements in data availability or quality that will then be used to increase costs Thus the market for new or improved measurement instruments or procedures will come from changes to regulatory or other controls; usually with long timescales and demonstration costs Both the users and the regulators become ‘fixated’ on a particular measurement practice or procedure when it has apparently ‘worked’ for a significant time An example is the use of ISO 7027 as the standard for turbidity instruments It has been well known for many years that polar plot of the scattered light gives much more information about the particles but ISO 7027 specifies a narrow beam and a specific wavelength A further example is the dependence on residual chlorine as an indicator of ‘safe’ drinking water in the distribution system when it is known that cryptosporidium is unaffected by residual chlorine Changing established standards

or guidelines can take many years even when the enthusiastic volunteers have been found and motivated

6.1.9 TECHNOLOGY

As already stated, in general the water and waste treatment and environmental indus-try is not limited by on-line measurement technology and it is likely that evolution of currently available techniques and products will be the dominant feature of market development

A wide range of techniques and technologies are used for on-line measurement

in the target industries including:

rvarious pressure techniques;

rvarious height measurements for open channel flow;

Technology 365

rmagnetic and ultrasonic techniques for closed pipe flow;

rpH;

rconductivity;

rion selective electrodes;

rchromatography;

rvarious light obscuration and scattering techniques for turbidity and particle analysis;

rvarious optical techniques to measure colour changes on paper strips or in cuvettes

or other transparent containers;

rvarious uses of spectral changes in the UV/IR/NIR/VIS regions;

rbiosensors;

rClarke cell;

rChemiluminescence.

There are four main technical drivers influencing developments for measurement technology: onboard intelligence; miniaturisation; serial communications, radio communications

It is expected that separation techniques and spectral analysis will become avail-able at the basic sensor level Chromatographs, ion mobility spectrometers and mass spectrometers are already deployed in the field but still need specialist support Data analysis is not yet robust enough for the casual user but, as shown below, the technology is available for this position to change

It is expected that highly capable portable and hand-held instruments will have

an increasing role, particularly when time, date, location data are automatically collected with each measurement Such instruments will have simple operator-free downloading facilities with operator identity thus providing more secure audit trails than is generally available in this industry

It is tempting to see an increased role for test kits but the need for secure audit trails needs to solved, if test kits are not to be limited in their use to investigative and process development roles

Elsewhere the argument is made that there will be regulatory and social pressures for many more measurements to be made with lower and lower detection limits, but this poses a capital expenditure problem However, there are technological drivers and enablers already being used and on the horizon which will permit autonomous sensors with much lower CAPEX and OPEX costs

Sensor development and commercialisation has been a multidiscipline activ-ity for decades but the increasing intensactiv-ity of the need for a multidiscipline ap-proach may well be a pacing issue The higher education institutions do not provide

multidiscipline courses and there is a notable reduction in networking and a reliance

on internet searching becomes the norm For example, blending microminiaturisa-tion with the latest optical, biosensor, battery and radio technologies is not a trivial exercise

6.1.9.1 Miniaturisation

Nanotechnology and microminiaturisation in its many and various forms has had huge publicity for a decade or more and is revolutionising sensors in other fields but the small scale of the water and waste treatment and environmental industries has hindered takeup However the massive amounts of resource being deployed for Homeland Security in the USA and elsewhere will provide useful modules and enabling techniques The pacing issue is the cost of the ‘foundries’ to ‘machine’ the structure; usually silicon but other materials, such as polymers, are beginning

to be explored We already know from the semiconductor industry that very large numbers of very low cost units are required to cover the overhead costs of a silicon foundry At the miniature sensor level there are some promising polymer ideas that might help parts of the market needs

An article in the November issue of Intech magazine (www.isa.org/intech) de-scribes microChemLab, developed by National Nuclear Security Administration’s Sandia National Laboratories The suitcase-sized equipment uses separation tech-niques and detection in the gas phase with a surface wave detector While it is not clear from the article what miniaturisation is involved the development direction is clear

A brief article in the February issue of The Engineer announced a miniaturised ion mobility spectrometer that will make hand-held detectors for lung diseases and airport explosives security While this is a gaseous phase measurement earlier phases

of this work have been shown to be useful for land remediation investigations using

a head space analysis approach

Even a brief web search finds a very large amount of activity in the many forms

on miniaturisation and many ‘clubs’

The idea of ‘smart pebbles’ – autonomous sensors that can be scattered over a battlefield and self network to discover ‘hot spots’ of some chemical or activity of interest – has been discussed for nearly a decade Outside of the military secrecy curtain they do not yet exist but there is now talk of ‘smart dust’

6.1.9.2 Battery

Battery technology already permits sensors with an installed life of 5 years and some of the more ambitious companies are claiming a 10-year installed cost This massively decreases the cost of installing sensors in areas where services are not already readily available