A model of accuracy

Since the launch of Water Innovate last year, the company has begun to release products from its portfolio of environmental technologies. The first of these was Odoursim, a software package that predicts the formation and emission of odour from different process components within a WwTW.

Wastewater treatment, waste management, agricultural and industrial practices have always produced unpleasant odours. But, in recent years, the number of complaints has increased.

This is happening for several reasons, including the encroachment of housing on land surrounding facilities, and the raised awareness of public rights in relation to environmental issues.

These growing environmental concerns have also led government bodies to consider policies to regulate the emission of odours, such as introducing boundary limitations. This has resulted in planning applications for the development or modification of existing sites being challenged – in the likelihood that a facility may have a negative effect on the local population.

This means that odour control and prevention has become a key issue both in the management of existing facilities and in the process of developing new sites to meet obligations under new environmental pressures.

Based on extensive research at Cranfield University’s Water Sciences group, Odoursim has been developed by Water Innovate to provide a design tool to help operators develop defensible odour management plans and abatement strategies for existing and planned installations.

The software uses wastewater characteristics that are convenient and inexpensive to measure – for example, flow, COD, soluble sulphide, pH and temperature levels – as input parameters, along with process sizes (physical dimensions and design parameters), and meteorological data available from sources such as the Met Office. The software then performs calculations that generate predictive odour emission readings in different areas of the treatment works, for instance, a trickling filter, weir or sedimentation tank.

The software has advantages over many currently used analytical methods, offering improvements compared with existing sampling techniques and the use of look-up tables (LUT), which can contain data that does not describe with sufficient accuracy the pattern of odour emissions actually observed from a site.

Importantly, Odoursim also allows the impact of variations in wastewater flow, quality and meteorological conditions on emission rates to be examined over time. The resulting dynamic emission rate data can then be used within dispersion models to produce a more representative odour contour plot rather than relying on an unreal assumption of a constant emission rate. The latter had been common practice until the development of this software package. The use of air-dispersion modelling systems (such as the US Environmental Protection Agency’s ISC and Aeromod) to predict odour concentrations originating from WwTWs is commonplace. The resulting odour contour plots are often used to support planning applications to show nuisance will be minimal. The models can also be used to make decisions about site-

specific odour control technologies.

Odour annoyance

However, it is essential that each element in the odour-annoyance pathway (formation, emission, dispersion, perception) is considered with equal importance (see Figure 1). An over-emphasis on dispersion modelling, where other pathway elements are treated simplistically, will give inaccurate predictions.

One of the main problems in modelling the pathway is the difficulty in measuring odours. Odorant concentration can be measured analytically, or the effect of odorants on the sense of smell can be evaluated using olfactometric methods.

The relationship between odorant concentration and perceived odours is complicated because sewage emissions contain many different odorants. However, hydrogen sulphide gas (H2S) is commonly used as an effective proxy indicator of overall odour strength because it is easily measured and it is one of the most common odorants associated with wastewater.

Using H2S, Odoursim concentrates on accurate mechanistic modelling of the formation and emission of odour. The software employs a COD-based liquid-phase H2S model and uses mass-transfer calculations for common wastewater process components such as weirs or sedimentation tanks.

Odoursim is compatible with existing dispersion models, which means that it is relatively simple to adopt. The software was recently used to develop a model of a medium-sized municipal WwTW in East Anglia, and to derive a 48-hour profile and 98 percentile yearly contour plots.

The WwTW had received a number of complaints concerning unpleasant odours over recent years as housing on the land surrounding the facilities increased.

There had been concerns over septicity issues at a raised inlet with a drop structure. Most recently, the primary sedimentation tank at the plant had been informally identified as the source of the odour. Previously, trickling filters, dating to the 1930s, were thought to be the potential source. And similar filter units, installed during 1970s, were also intermittently causing issues.

Wastewater sampling

Wastewater sampling for total COD, soluble COD, sulphide, pH and temperature was carried out at the plant over a 48-hour period to obtain composite samples on a two-hourly basis.

In addition, micro-meteorological sampling – H2S mapping, using a Jerome gas analyser – was done over a one-hour period on each of the wastewater sampling days, and emission rates were back-fitted using the Aeromod model and H2S mapping results.

An Odoursim input file, covering 48 hours was created from diurnal flow and load pattern sampling, and a model of the site was built using process sizing information (see Figure 2). The liquid-phase of the model was then calibrated to the measured wastewater characteristic trends over the first 24 hours. The gas-phase of the model was calibrated to the emission rates derived from the first day of the micro-meteorological study.

The Odoursim model was then run over the full 48-hour period and the emission rates from the identified sources were compared with a second micro-meteorological study, to ensure validation of the model over a second 24-hour period.

The results of the calibrated and validated 48-hour Odoursim model run were used as inputs for the Aeromod model, and 48 one-hour runs were performed to yield the emission contour plots for each hour of the 48-hour period. A wastewater input file was generated using the diurnal flow and load pattern, and the wastewater influent flow in 2004. The calibrated Odoursim model was then run for the 2004 period using the new wastewater input file.

The results from the Odoursim model were then entered into the Aeromod model to yield the 98 percentile contour plot for the 2004 period. A similar Aeromod model was run using standard constant emission rates to yield the 98 percentile contour plot for the 2004 period.

A good fit between the calibrated Odoursim model and the micro-meteorological measurements taken on site was observed. This indicates that the biological, chemical and physical equations used in the software model accurately describe emissions from the plant for a given point in time. The model also accurately predicted the emission rates for the identified sources for the validation measurements.

Traditional methods

Odoursim reveals that traditional methods, for example, using values obtained from an LUT, do not adequately represent the pattern of emissions observed from the facility. In fact, a comparison of 98 percentile contours for both the relevant Odoursim (see Figure 3) and LUT (see Figure 4) modelling shows that the odour source was incorrectly identified over a period of 12 months.

The study reveals that the new trickling filters – installed 30 years ago – are the main source of the odour and not the old units. The time-slice contour maps that were generated showed how diurnal patterns of emissions from the raised inlet drop structure are linked to flow and load. This presents opportunities to model return liquor load impacts on downstream septicity.

Also, by showing that the raised inlet drop structure was responsible for high emissions (see Figure 5), the model is able to provide the process design team with design loading data for odour control equipment, and help size the correct solution to this problem.

Current methods and LUTs do not predict dispersion patterns with regards to measured values on-site. These techniques are only good for average conditions; they do not reveal diurnal patterns, unless excessive sampling is used; and they provide only a snapshot, which has no predictive capabilities. Odoursim is intended to complement existing odour dispersion models, and improve the accuracy and validity of the final output by providing a precise and mechanistic method for inputting odour formation and emission data.

Because the software is compatible with existing dispersion models, its adoption is relatively simple. The improved level of confidence from dispersion modelling exercises will be important to water utilities, plant operators and regulatory authorities alike in the cost-effective management of odour abatement and public perception issues.

Odoursim has a number of benefits. The modelling software:

• Offers improved accuracy in contour plots – using dynamic emission rates
• Provides greater confidence in predictive abilities of dispersion models
• Enables process changes to be evaluated in refurbishment scenarios
• Models flow and load changes for future design load scenarios
• Determines peak-to-mean ratios and statistical spreads of emission rate from various process sources