Advertisement

TSG Chile



PrOlor, Forecast 2 Days Beforehand an Odour Incident

on . . Bisitak: 9927

ProOlor. Prognosis of Odours

proolor software   There is a wide range of methods used to monitor incidents at a real time scale using dispersion models in some occasions coupled with electronic sensor gas emission monitoring. In other cases the monitoring is made with the help of real data from residents or with dedicated panellists. They all have one thing in common: These tools predict the event when it is already happening, therefore there is no option for the plant manager to take actions to correct the incident.

David Cartelle Fernándezb, Carlos N. Díaz Jiménez*a, Jose M. Vellón Grañab, Ángel Rodríguez Lópezb.

a SVPA, Servicios de Protección Ambiental

b TROPOSFERA

 Contacto:

website: www.prolor.net

  

There is a wide range of methods used to monitor incidents at a real time scale using dispersion models in some occasions coupled with electronic sensor gas emission monitoring. In other cases the monitoring is made with the help of real data from residents or with dedicated panellists. They all have one thing in common: These tools predict the event when it is already happening, therefore there is no option for the plant manager to take actions to correct the incident. In this paper we present PrOlor, a novel tool to predict odour emission episodes with up to a week beforehand, using prediction meteorological data instead of real time data. This way the plant operator have a time frame to take actions to control the odour emission and prevent odour incidents, before they actually happen.

1. Introduction

    Dispersion of atmospheric pollutants modelling is today a routine method in environmental air quality management. This is so because of the difficulty of obtaining a reliable value of odour concentration at the receptor.

   Nowadays, it is possible to predict the odour impact of an activity by using odour emission rates at the source coupled with dispersion modelling.

   As the calculation capacity of the computers grew, it was possible to calculate and see the odour plume in nearly real-time with the help of a weather station. This made possible to the industrial activities to see in a screen when there was an impact in the neighbourhood at the same time that it was happening.

   A further improvement was reached when it was possible to install a sensor/group of sensors to measure a set of patterns in a source. These patterns were linked to a chemical compound or in some cases to dynamic olfactometry measurements. This way it was possible to further study the emission of gases in the activities with the information of the emission points in a real-time scale.

   However, all these systems share a problem in common: In most of the cases, there is no way to prevent the odour impact because it is linked to a release of odours from the source that happened from minutes to hours beforehand! Once the impact is observed at a real time, it is already too late to take actions to correct it.

   This is important because many industrial activities can control their emission by means of operative actions, such as a decrease or delay in the production of a unit or an increase of the efficiency of the odour abatement system. This can be done for example by using more chemicals or maybe increasing the speed of fan to favour the dispersion of the plume. If the industrial operator knows well in advance when, where and which is going to be the odour impact at the source, operative actions can be taken in order to prevent it.

   Nowadays most of the smartphones have applications that shows the weather forecast in a certain location. This software is based on predicted meteorological data and in some cases are used as quick and inexpensive tool for the plant operator in order to take operational actions, for example if the wind direction is expected to be toward the population. However, there is no way to predict which will be the odour concentration at a certain location at a certain hour.

   The aim of this paper is to present PrOlor, a novel software for smartphones and computers, specifically designed to forecast the odour impact of an activity to up to 72 hours beforehand, using prediction meteorological data instead of real time data.

2. Materials and methods

   PrOlor is based on the WRF/CALMET/CALPUFF model system, built and installed for Linux platforms.

   The CALPUFF model (Scire et al., 2000) is a Lagrangian, multi-layer, multi-species, non-steady state puff model, used for the diagnosis of the dispersion of pollutants in the atmosphere. CALPUFF is suited to model odours (Barclay et. al., 2013; Diaz et al., 2014).

   The CALPUFF dispersion system consists of 2 main modules: The meteorological diagnostic model CALMET and the air dispersion model CALPUFF.

   The CALMET module uses surface and upper stations (radiosounding) or other meteorological models. The advantage of this meteorological model over traditional Gaussian solutions (e.g. using a single surface meteorological station) is noticeable, since it is able to simulate conditions at a local level that completely change the meteorological scenario, and therefore the dispersion of pollutants.

   To feed the CALMET model with meteorological data, the WRF meteorological model (Weather Research and Forecasting) mesoscale is first executed.

   WRF (http://www.wrf-model.org) is a modern meteorological generation model that gives wind fields, pressure, temperature and humidity data with high spatio-temporal resolution, which are very important as entry data for air quality models. The WRF model has the peculiarity of being locally configured to represent spatial domains at different scales according to the study that is to be done.

   In PrOlor, the WRF model is executed daily for a time horizon of 48h, initialised from GFS data from the National Centers for Environmental Prediction (NCEP). Using FNL synoptic scale conditions, a pattern of nested domains is followed until a modelling domain at high resolution is obtained obtaining hourly data of over 20 meteorological parameters and at 27 different height levels.

   The map is then configured with the two nested domains (D1, D2, and D3 of 27, 9 and 3 km2 resolution, respectively) used in meteorological forecasting.

   WRF is designed for use in both prediction functions and reanalysis. It has a modular architecture, allowing different parameterisations of a dynamic or physical nature to be applied, among others. It also offers various assimilation systems with actual data, in addition to a software development paradigm that allows its execution in both personal computers and large parallel computing stations. WRF is suitable for a wide range of applications at different scales and can operate at resolutions of hundreds of metres to thousands of kilometres.

   Once the WRF process has been completed, the CALMET 3D model for the forecast period is executed using the following methodology:

  1. The WRF output file is processed by the CALWRF routine that is responsible for reading this output and transforming it into a 3D.dat file, a format that is accepted for the CALMET model input.
  2. This file contains all the height and surface meteorological data that the WRF model contains.
  3. The meteorological downscalling is done in CALMET - an increase in the WRF output model resolution - up to 200 m - and subsequent to that, the CALPUFF model is executed.

   The input of the CALMET model through the WRF model is a substantial improvement because this provides much more information on surface and height.

diagram

Figure 1: Input data needed to feed the CALPUFF modeling software.

   In PrOlor 5 height levels are modelled at 20, 50, 300, 1,000 and 3,000 m.

   The emission rates in ouE/s are calculated. These emission rates can be either obtained by emission measurements at the source with dynamic olfactometry or they can be estimated with emission factors.

   Finally to correctly analyse dispersion, the CALPUFF model integrates a terrain module. In this module, topographic data are constructed from land use data as well as a digital terrain model of the study area.

   In a normal basis, topographic data is obtained in the SRTM3 mission (Shuttle Radar Topography Mission), whose resolution is approximately 90 m. However other topographic data with a higher resolution (such as LIDAR) have been also successfully tested.

   Land use data is obtained from the GLCC (Global Land Cover Characterization). These data have a resolution of 300 m and contain 22 kinds of land uses defined according to the LCSS (Land Cover Classification System). The GLCC is based on observations made by the MERIS sensor from the ENVISAT satellite between December 2004 and June 2006.

   After this, the data are fed to CALPUFF and the software is run. The operation data in netCDF format is used for graphing, creating data files, generating kml files (Google Earth) and archiving "in the cloud".

   The results of the odour modelling can be accessed through web and mobile apps (Android).

3. Results and discussion

   The PrOlor system is currently used in a paper pulp factory. In this particular case there has been a modelization of both odour and chemical concentration. Everyday there is a new run and the operator can detect two days beforehand if there is going to be an impact in the population nearby. According to this predictions, the operator can take actions to try to minimize the concentration of chemicals or in some cases maximize the flow at the stack so that the dispersion is increased.

   The paper pulp factory can delay the production in some critical hours to minimize the odour emission at those hours.

   Furthermore, the operator can comfortably check in an android smartphone the prevision data in a software similar to that of the web. Therefore, there is no need to stay close to the office in order to take actions to minimize the impact of the odour emission on the receptors.

   The system can be coupled to online measurement systems to further increase the accuracy of the results obtained.

ProOlor

Figure 2: Output of the CALPUFF modeling software.

   It is difficult to check the effectiveness of the method, as it is based on actions that change the course of the future. To check compliance, it is already planned to use this software in a rendering plant with a previous register of complaints, so that it will be possible to compare these data with the ones of the previous year. This way it will be possible to check if there is any change on the number of complaints.

   In addition it is planned to use peak to mean ratios to further improve the output of the CALPUFF modelling software (Diaz C. 2009).

   This experiment will be carried out for a year starting in the summer of 2014. it is expected that some preliminary data will be released in the NOSE conference 2014.

4. Conclusions

   The online odour modelling software available in the market now is able to check the extension of a plume and its impact in the receptors. In this way, an impact can be observed but there is no time to take actions beforehand.

   Taking a proactive approach is the best way to save time, money and keep the industrial activity in compliance with the regulators.

   If a facility have the flexibility to delay or modify their production units in order to increase production when there is no odour impact and decrease it when there is certain impact, then a software like PrOlor that it is able to predict a couple of days beforehand the impact in a receptor, is a very useful tool.

References

Bächlin, W.; Rühling, A.; Lohmeyer, A., 2002. Bereitstellung von validierungs-daten für geruchsausbreitungs-modelle-naturmessungen., Baden-Württemberg Environmental Agency.

Barclay J., Borissova M., 2013, Potential Problems Using Aermod To Implement Current Odor Regulations For WWTPs, 5th IWA conference on Odours and Air Emissions, San Francisco, USA.

Diaz C., 2009, The Fascinating Study of the Peak to Mean Ratio <www.olores.org> accessed 11.07.2014 (in Spanish).

Diaz C., Cartelle D., Barclay J., 2014, Revision of Regulatory Dispersion Models, an Important Key in Environmental Odour Management, Ist International Seminar of Odours in the Environment, Santiago, Chile.

Capelli L., Dentoni L., Sironi S., Guillot JM, 2012, Experimental Approach for the Validation of Odour Dispersion Modelling, NOSE 2012.

German Ministry of Environment, Technische Anleitung zur Reinhaltung der Luft, 2002

Scire J., Strimaitis David G., Yamartino Robert J.2000. A User’s Guide for the CALPUFF Dispersion Model

  • No comments found
Add comment