Testing the water

Monitoring the quality of water and wastewater can present some significant technical challenges for utilities and manufacturers as well as for suppliers of process equipment and analytical devices.

The Pontsticill water treatment works (WTW) in the Brecon Beacons National Park, for instance, was using bellows pump technology for its turbidity testing.

The inherent introduction of air in the samples, however, led to false readings - a problem that has been since resolved by a recent switch to a new pump technology.

The D?r Cymru Welsh Water facility draws its water from the Pontsticill Reservoir, holding up to 3400 million gallons of water for supply in South Wales. To improve its turbidity testing, the Welsh utility decided to replace one of its bellows pumps with a new Qdos 30 metering pump from Watson-Marlow.

The Qdos 30 metering pump at Pontsticill takes elutriant from a DAF (dissolved air flotation) filtration plant and lifts it about three metres before it passes through a turbidity meter and returns by gravity back into the flow.

“This is a clean water site and the quality of sample is very important,” said Marek Cegielski, a process scientist and manager at D?r Cymru Welsh Water.

“In the past we’ve tried other types of turbidity monitoring devices, such as dip probes, but without much success. We’ve even used bubble traps in conjunction with our bellows pumps, but we still fell short of 100% reliability.”

“Our main requirement is for a constant, pulse-free flow,” said Cegielski. “Previously we had tried fitting back valves to bellows pumps to stop the pulsing, but in truth we couldn’t properly buffer the effect and still ended up with false readings.”

Installed in July 2012, the Qdos 30 is working well without any signs of air introduction or flow variation, according to Cegielski: “So far, so good, as far as we’re concerned…In total we have four streams to monitor at Pontsticill, so if our existing Qdos 30 continues to meet expectations then we’ll place an order for a further three pumps.”

A different type of challenge emerged at an unnamed water company after its installation of in-pipe drinking water quality monitoring sondes. Distribution engineers discovered an occasional significant disparity between the data from water monitoring devices and results from grab samples tested by a photometer tablet (DPD) method.

For most of the time, there was close agreement between the two methods, but periodically, the photometer results varied wildly, whilst the in-pipe monitors showed relatively stable values.

Initially, distribution managers suspected that the online monitors were at fault and that they were failing to detect occasional spikes in free chlorine. However, as a multi-parameter monitor, the Intellisonde was able to provide more clues which ultimately lead to the identification of the problem.

The devices provide real-time monitoring data for water within pipeline distribution networks. They feature up to 12 sensors inside a tiny sonde head that fits in a water pipe through a 3.8 or 5cm valve.

The water supply in this study was from a desalination plant. Infrastructure was fairly new, and the water company took water from a processor and a trunk distributor.

To check the performance of the devices, grab samples were taken and free chlorine tests were conducted with a portable photometer which measured the colour intensity produced when DPD tablets were crushed and mixed with a sample.

In an initial study, regular samples were taken manually over a 24 hour period and tested with the DPD method. These results were compared with the Intellisondes’ logged free chlorine data. An unexpected scatter of DPD measurements was observed during night time and on multiple occasions.

The data showed that at the time when the DPD measurements were most scattered; flow dropped and turbidity increased, so it was suspected that the particles contributing to this turbidity could be interfering with the DPD measurement.

The process of sampling caused flow disturbance in the pipe which appeared to result in a variable level of particles in each sample; low flows at night were resulting in the settlement of particles which were disturbed (randomly) by the sampling process, and it was considered likely that these particles were affecting the DPD results.

To check this assumption, local sand was added to bottled drinking water and tested by the DPD method. This water had previously been tested and shown not to contain free chlorine, but following the addition of local sand the DPD test gave a positive result for free chlorine.

The reagents in DPD tablets are known to react with a wide range of species, including iron and manganese compounds, and it was discovered that local sand grains have a coating of iron oxide which gives them a pink colour. It is highly likely therefore, that this iron oxide coating was reacting with the DPD reagent and causing the false free chlorine measurements.

“This project highlights the benefits of multiparameter monitoring - without the flow and turbidity data, we might never have surmised that the DPD results were incorrect and that sand was the cause,” said Jo Cooper of Intellitect Water.