Measuring Ammonia accurately for the Environment Act 2021

Many parameters in monitoring water quality can be measured relatively easily using very well-established techniques that range from physio-chemical and optical sensors through to reagent-based analysers. There are few parameters in water quality monitoring that provoke such a wide response of opinion as with ammonia measurement. There are a myriad of suppliers and sensors that offer seemingly unique characteristics and performance advantages. For the Environment Act 2021 and the WINEP programme, 5 key determinants have been highlighted by DEFRA: turbidity, pH, dissolved oxygen, temperature and ammonia. In order to measure these determinants, multiparameter probes or sondes have been quickly highlighted as the most obvious instrument to monitor these parameters based on flexibility, performance and price. In reality any instrument that can meet proposed DEFRA performance specification should be considered; multiparameter probes are simply one instrument of many that could be used successfully to monitor the impact of Combined Storm Overflows (CSOs).  Multiparameter sondes have been designed to bridge the gap between high end and expensive single parameter analysers and simply taking samples to laboratories or taking spot samples with simple hand-held meters. 

Outright Need for Innovation

However, many water quality determinants have been neglected due to a lack of innovation and have typically utilised ageing measurement techniques. One such parameter is ammonia (NH3) which utilises an ammonium ISE (NH4) to measure ammonium salts in the water. The principle behind ISEs is that a selective membrane allows specific target ions to pass through a porous membrane and then measured by an electrode inside the sensor. Once the pH of the water rises above ~pH7.5, ammonium salts begin to convert to ammonia gas (which can be very detrimental to aquatic ecosystems especially when pH rises above 8.5). There is a point of inflection at around pH9.2 where most of the ammonium in the water will be converted to ammonia; by pH11 almost all ammonium will be converted to ammonia. pH and temperature correction are essential to determine this relationship before providing an output but not necessarily as important as pollution indicators in their own right. It can therefore be argued that ammonia is probably the single most important water quality parameter under the Environment Act. Dissolved oxygen level reactions can take time and don’t necessarily occur at the point of pollution but instead represented by a lag which could be several days down river; therefore, its use to specifically target and quantify pollution sources is somewhat limited.
Ion-selective electrodes are often selected due to their relatively inexpensive cost at point of purchase, but this is very much a false economy. The membrane-based sensors have a limited operational lifetime before the membrane expires and as a technique the membrane remains vulnerable to interference from other ions and potential drift. While ammonia forms an important part of the picture, it is questionable as to whether with the new monitoring requirements, ammonium ISEs are suited to being the main parameter by which pollution signals are tracked in water. By comparison, using fluorescence to measure BOD or ammonia, has less interference and uses optical technology that has a long lifetime and low maintenance cost.
It is therefore essential that we understand the drawbacks of using ISEs before utilising them as a reliable form of measurement; after all, if they are not accurate enough for the purposes of the Environment Act then it can be argued that the billions of pounds that will be spent, will be simply wasted causing water companies, the public, environmental enforcement agencies and the government more problems than solutions. Ultimately, we need reliable sensors that allow us to make competent and informed decisions, ones that will improve our environment; bad data could have the opposite effect. Let’s consider some of the issues of ISEs:

1. Poor accuracy
Almost all ISEs (not installed within analysers) have a typical accuracy of ±2mgl. Using an ISE to accurately depict pollution is a temperamental process and there have been many references by enforcement agencies that they can only really be used as a trend notification rather than absolute measurement. To make things worse most forms of pollution (whether its wastewater, agricultural or industrial) would effectively be diluted below the accuracy attainable by an ISE. By using table 1 below, the data demonstrates how CSO discharges into a receiving water course (based on a 10mgl NH4 discharge) most scenarios of pollution would be within the red triangle. It is widely recognised that no ISE can provide this level of accuracy so clearly, we need innovation to help! What makes things worse is that ammonia levels would then need to be calculated utilising pH/temperature correction, therefore exacerbating the potential accuracies even further!
2. Drift
All ISEs will drift, and it is unfair to allow any manufacturer to let any user assume that they will not and that they are stable. Furthermore, the exact drift pattern has until recently not been understood and it has been assumed that ISEs drift gradually. This is in fact not true, and the drift is often either triggered by an event or accelerated by it. Any attempts to re-process the data by applying a drift factor would not be recommended.
3. Maintenance
ISEs require significant maintenance which involves calibrating typically every 4 weeks and replacing membranes approximately every 6 months. This is a very costly and time-consuming exercise and with 20000+ installations proposed under the EA21 this is pretty much a resourcing and carbon nightmare for users.
4. Failure
Even with significant maintenance regimes, ISEs have a propensity to fail which will instigate further call outs and costs. There are no manufacturers currently offering warranties greater than 12 months for their ISEs within sondes, and for very good reasons!
5. Monitoring Levels
A healthy river should have ammonium levels of below 0.2mgl which requires an ISE to work accurately and reliably within the <0.5% of its full-scale range. Not many sensors can do that including ISEs! Then add its background accuracy of ±2mgl.
6. Data Integrity
With all the issues that have been highlighted, a resolution of 0.01mgl and an accuracy of ±2mgl it is hard to believe how an ISE could ever provide accurate and defensible data for the EA21. If an organisation was being prosecuted for a 1mgl spike in ammonium/ammonia, then it could obviously be argued that reading could be anywhere between -1 and 3mgl. We simply need a better and more reliable approach to measuring ammonia!
So that’s all the bad news. Is there any good news? In fact, there is and quite a lot; DEFRA are actively encouraging water companies to innovate and although the technical guidance released for the EA21 admits that ammonia measurement needs to be improved, DEFRA have released a more recent update (August 2023) which requests that water companies should utilise multiparameter instruments with 2 spare parameter ports to allow for future innovation. There is without doubt a massive requirement to provide more innovative water quality measurement. At this moment in time Proteus Instruments are the only company that can provide these extra 2 parameter ports; in fact, we can offer 6 spare ports! Furthermore, at the time of writing this, DEFRA/Innovate UK have released a funding call for innovation projects to improvement environmental monitoring instrumentation. This should also encourage water companies to adopt more innovative monitoring instrumentation for the purposes of the EA21.
Although we at Proteus recognise that there are potentially more valid water quality measurement parameters than ammonia which include BOD, COD, TOC, E. coli and even phosphorous, we do acknowledge ammonium/ammonia is here to stay and as such this measurement technique simply needs significant improvement. As it stands, the OPEX burden of maintaining 20,000+ instruments with ammonium ISEs are mind-boggling and almost unjustifiable. Simple maths would indicate that 22,000 instruments would require a minimum of 12 visits per annum and assuming 4 instruments per day (per field team) that’s 5000 man-days per month, and that’s assuming no callouts. There is simply not the resource in the UK to achieve this, let alone the carbon footprint that it would create. Ultimately, we have to remember that this cost would be passed on to the consumer and our environment. So, this is clearly a situation that nobody is happy with except perhaps any potentially unscrupulous manufacturers or persons. The morally right thing is to innovate and provide a solution that is more accurate, improves reliability, reduces costs, builds trust with the public, minimises resources and reduces the carbon footprint.
Proteus are ahead of the game and have invested heavily in innovation since it was formed in 2018. With multiple world firsts, a Queens Award for Innovation (2022) and pivotal contract wins, Proteus have now also released the world’s first optical instantaneous ammonium measurement technique for a sonde. Utilising a Proteus with multiple fluorometers, Proteus are able to provide high accuracy, highly stable ammonium measurement. Furthermore, the sensors will only require calibration once per annum saving water companies 90-95%% of their OPEX budget that would be associated with calibrating and maintaining ammonium. Overall, in any one year we estimate this single technological improvement will save a water company’s OPEX by 65%. Table 2 below demonstrates a typical calibration calendar for a multiparameter sonde. As you can see 12 visits are required for any NH4 ISE whereas only one visit is needed for optical NH4. If the move is made to optical it would ultimately mean that probes would only need 3-monthly visits due to pH. Hence, only 1 in 3 visits would be required using optical NH4 technology from Proteus. There is currently technology that would push out pH calibrations even further which has the potential to instigate a 6- or 12month calibration run. The cost savings for this are significant but do not compare to the savings associated with extending NH4 calibrations from 1-month to 12-months.
So, let’s see just how much it would cost to run an optical system versus a conventional ISE system. We have utilised conservative costs based on an in-river solution; a kiosk-based system is likely to cost c.30% more in terms of CAPEX and also an increase in OPEX due to extra maintenance. Labour costs have been approximated to £250 per visit and each ISE based system requiring 2 replacement membrane caps per annum at £300ea. The table below simply demonstrates the savings that can be attained by utilising optical technology based on 20,000 installed units in England:
Just by switching to an optical based system an estimated £420,000,000 could be saved within 10 years; that’s over £45million per water company or a 41% total life cost saving! This does not include any extra call outs that would be required to attend drifting or failed ISE sensors. Furthermore, it does not take into account the massive carbon saving associated with having to have 3 times more site visits than with an optical system. This aspect alone could in some minds even outweigh the actual financial savings of an optical system even further! What is obvious is that the operational costs far outweigh the purchase costs by 3:1 whereas the optical systems are only 1.5:1. The savings are quite stark and clearly questions why we are even looking to use sensors which are not designed for long-term performance. This fundamentally backs up DEFRA’s concerns of using ISEs and hence the call for innovation to revolutionise water quality monitoring; something we have been doing since 2018. What is particularly exciting is that this is only really the start; it is hoped that innovation and technology will provide even more accurate and reliable forms of measurement, reducing costs even further. One thing is for sure, we must embrace innovation and not stick to out-dated forms of measurement.

How good is Optical?

So, for optical measurement, just how good is it…below is a calibration curve showing 1988 stable ammonium readings (proven by interim calibrations) versus the Proteus’s optical output. Clearly it is possible to see that the correlation is incredibly robust although we always recommend undertaking a local site calibration to confirm or optimise the readings. This can be done effortlessly by utilising a good quality ammonium ISE to record data simultaneously for a couple of days. The Proteus then learns and provides an optical measurement utilising multiple fluorescence sensors and a turbidity sensor. Unlike an ISE, the sensors then only need to be calibrated once per year. Although other sensors such as pH will need 3-monthly calibrations, this reduces the number of site visits required by 66%. Furthermore, replacement membranes and associated callouts will be a thing of the past.
Below is a graph showing 6 weeks of continuous 15-min data. Shortly after calibration an event triggers a very small amount of drift which is maintained throughout the deployment. Upon calibration (11th October) the ISE is recalibrated and completely agrees with the optical ammonium sensor. Although these are relatively low levels of ammonium in this example, typical drift is often far greater. This data really exemplifies the accuracy and repeatability that optical measurement can bring.
 Figure 1: NH4 ISE vs Optical Dataset with minor ISE drift

With optical measurement water companies and regulators will get accurate, more meaningful and more defensible data. More importantly optical ammonium will provide water companies with a reliable means to achieve the aims of the EA21 but it also comes with other great advantages; (1) sensor life is in excess of 10 years unlike ISEs (<5years); (2) optical ammonium can measure significantly lower than ISEs; (3) can be used in environments where ISEs cannot be used such as marine or more saline waters; and (4) are not prone to the interferents that ISEs are acknowledged with.
There are clearly a whole multitude of reasons to use optical measurement over ISE-based measurement but if we are to be realistic about achieving the aims of the EA21 then we need to innovate now. Installing 22000 ISE sensors from any manufacturer(s) would be the equivalent of driving towards a cliff edge at a rate of knots. As yet we have not mentioned anything regarding standards of installation or calibration; there is also a fundamental need to provide repeatability across installations and calibrations. Although there is some discussion about this there will need to be some degree of flexibility in this but ultimately will need a consultation with all manufacturers to incorporate best practice in a national installation and calibration standard. If we don’t, then when the monitored networks get too big to manage and the data quality is so poor, there will ultimately be a significant backlash from water companies, regulators and the public questioning how is this sustainable or in fact trustworthy. With such an obvious outcome it is necessary to innovate now rather than later. Right now, there are great solutions available that will provide better and more cost-effective data. Those who adopt innovation, will reap the rewards by having maintable water quality networks with good defensible data; those who don’t will have networks that will descend into chaos and without doubt have significant interactions with environmental regulators, shareholders and the public.