Reconstruction of the Chemical Gas Concentration Distribution Using Partial Convolution-Based Image Inpainting.

An interpolation method, which estimates unknown values with constrained information, is based on mathematical calculations. In this study, we addressed interpolation from an image-based perspective and expanded the use of image inpainting to estimate values at unknown points. When chemical gas is dispersed through a chemical attack or terrorism, it is possible to determine the concentration of the gas at each location by utilizing the deployed sensors. By interpolating the concentrations, we can obtain the contours of gas concentration. Accurately distinguishing the contours of a contaminated region from a map enables the optimal response to minimize damage. However, areas with an insufficient number of sensors have less accurate contours than other areas. In order to achieve more accurate contour data, an image inpainting-based method is proposed to enhance reliability by erasing and reconstructing low-accuracy areas in the contour. Partial convolution is used as the machine learning approach for image-inpainting, with the modified loss function for optimization. In order to train the model, we developed a gas diffusion simulation model and generated a gas concentration contour dataset comprising 100,000 contour images. The results of the model were compared to those of Kriging interpolation, one of the conventional spatial interpolation methods, finally demonstrating 13.21% higher accuracy. This suggests that interpolation from an image-based perspective can achieve higher accuracy than numerical interpolation on well-trained data. The proposed method was validated using gas concentration contour data from the verified gas dispersion modeling software Nuclear Biological Chemical Reporting And Modeling System (NBC_RAMS), which was developed by the Agency for Defense Development, South Korea.

Methane emissions from five Danish pig farms: Mitigation strategies and inventory estimated emissions.

This study investigated whole-farm methane emissions from five Danish pig farms with different manure management practices and compared measured emission rates to international and national greenhouse gas inventory emission models. Methane emissions were quantified by using the tracer gas dispersion method. Farms were measured between five and eight times throughout a whole year. One of the farms housed sows and weaners (P1) and the others focused on fattening pigs (P2-P5). The farms had different manure treatment practices including biogasification (P3), acidification (P4-P5) and no manure treatment (liquid slurry) (P1-P2). Quantified methane emissions ranged from 0.2 to 20 kg/h and the highest rates were seen at the farms with fattening pigs and with no manure treatment (P2), while the lowest emissions were detected at farms with manure acidification (P4 and P5). Average methane emission factors (EFs), normalised based on livestock units, were 14 ± 6, 18 ± 9, 8 ± 7, 2 ± 1 and 1 ± 1 g/LU/h, for P1, P2, P3, P4 and P5, respectively. Emissions from fattening pig farms with biogasification (P3) and acidification (P4-P5) facilities were 55% and 91-93% lower, respectively, than from farm with no manure treatment (P2). Inventory models underestimated farm-measured methane emissions on average by 51%, across all models and farms, with the Danish model performing the worst (underestimation of 64%). A revision of model parameters related to manure emissions, such as the estimation of volatile solids excreted and methane conversion factor parameters, could improve model output, although more data needs to be collected to strengthen the conclusions. As one of the first studies assessing whole-pig farm emissions, the results showed the potential of the applied measuring method to identify mitigation strategy efficiencies and highlighted the necessity to investigate inventory model accuracy.

A Simulation Framework for the Integration of Artificial Olfaction into Multi-Sensor Mobile Robots.

The simulation of how a gas disperses in a environment is a necessary asset for the development of olfaction-based autonomous agents. A variety of simulators already exist for this purpose, but none of them allows for a sufficiently convenient integration with other types of sensing (such as vision), which hinders the development of advanced, multi-sensor olfactory robotics applications. In this work, we present a framework for the simulation of gas dispersal and sensing alongside vision by integrating GADEN, a state-of-the-art Gas Dispersion Simulator, with the Unity 3D, a video game development engine that is used in many different areas of research and helps with the creation of visually realistic, complex environments. We discuss the motivation for the development of this tool, describe its characteristics, and present some potential use cases that are based on cutting-edge research in the field of olfactory robotics.

Numerical simulation of gas dispersion from rooftop stacks on buildings in urban environments under changes in atmospheric thermal stability.

The prediction of dispersion of gases emitted from rooftop stacks in a built environment is important for preventing or minimizing their harmful effects on human health. In this study, the wind flow and dispersion of exhaust gas emitted from rooftop stacks on buildings in an urban environment under different atmospheric thermal stabilities were investigated using numerical simulations. The wind flow field and dispersion contaminants were simulated using a computational fluid dynamics model with the k-ε turbulent schemes being resolved by the Reynolds-averaged Navier-Stokes approach. An isolated building was modeled under conditions of varying thermal stratification of the boundary layers (neutral, unstable, and stable conditions). The diffusion flow field within the building wake zone was investigated for various stack sites (center, right side, and left side). Experiments were conducted in a wind tunnel to validate the numerical simulation results, by using the data qualitatively and quantitatively. The numerical simulation results were consistent with the experimental observations. The results indicated that the pollutant concentration of the plume spread was high near the stack and decreased with increasing distance from the stack. Under stable conditions, the flow motion and separation increased in the wake zone, and the pollutant concentration of the lateral spread at the average human height decreased. Under unstable conditions, the flow of the vortex circulation was fast and strong, and the pollutant concentration of the vertical spread was high.

Assessment using CFD of the wind direction on the air discharges caused by natural ventilation of a poultry house.

Air inside poultry houses must be removed on a regular basis to prevent excess of heat, particles and noxious gases that can imperil animals. To cope with this issue, natural ventilation could be an effective method when assisted by accurate predictions. This study investigates air discharges caused by natural ventilation of a poultry house by means of a three-dimensional computational fluid dynamics (CFD) model. It solves the governing equations of momentum, heat and mass transport, radiative transfers and animal-generated heat. Wind directions of 0°, 36° and 56° (0° corresponds to a wind blowing perpendicular to the ridgeline) were investigated; the CFD model predictions achieved a RMSE of 1.2 °C and 0.6 g[H2O] kg-1 [dry air] for internal temperature and absolute humidity, respectively, when air blew with an angle of 36°. Air renewal rates (ARR) were 39.5 (± 1.9), 34.9 (± 2.2) and 33.6 (± 1.7) volumes of the building per hour, when air blew at 0°, 36° and 56°, respectively. Such ARR predictions served to know how the gases contained in air would likely spread downstream from the building in order to define regions of potentially high gas concentration that could endanger neighbouring habitable facilities.

Methane emission rates averaged over a year from ten farm-scale manure storage tanks.

Methane (CH4) emissions from animal manure stored in outdoor tanks are difficult to predict because of several influencing factors. In this study, the tracer gas dispersion method (TDM) was used to quantify CH4 emissions from ten manure storage tanks, along with the collection of supporting information, in order to identify its emission drivers. The dataset included two tanks storing dairy cattle manure, six holding pig manure, and two with digestate from manure-based biogas plants. CH4 emissions from the tanks were measured six to 14 times over a year. Emissions varied from 0.02 to 14.30 kg h-1, or when normalised by the volume of manure stored, emission factors (EFs) varied from 0.05 to 11 g m-3 h-1. Annual average CH4 EFs varied greatly between the tanks, ranging from 0.20 to 2.75 g m-3 h-1. Normalised EFs are similar to literature values for cattle and digested manure, but at the high end of the interval for pig manure. The averaged manure temperature for all tanks varied from 10.6 to 16.4 °C, which was higher than reported in a previous Danish study. Volatile solids (VS) concentration was in average higher for cattle manure (ranging from 3.1 and 4.4 %) than pig manure (ranging from 1.0 to 3.6 %). CH4 emission rates were positively correlated with manure temperature, whereas this was not the case for VS concentration. Annual average EFs were higher for pig than for cattle manure (a factor of 2.5), which was greater than digested manure emissions (a factor of 1.2). For the pig manure storage tanks, CH4 emissions were higher for covered tanks than for uncovered tanks (by a factor of 2.3). In this study, manure storage tanks showed a large disparity in emission rates, driven not only by physical factors, but also by farm management practices.

Quantification of - And determining factors affecting - Methane emissions from composting plants.

Methane is a potent greenhouse gas contributing to climate change. Reliable data for methane emissions from the waste management sector are paramount in terms of providing national methane budgets and developing climate mitigation efforts. This study quantified total methane emissions and characterised temporal as well as operational emission patterns at five commercial composting plants in Denmark. Methane emissions were measured over a one-year period, using the tracer gas dispersion method. The results show that methane emission rates ranged from 8.0 ± 0.1 to 42.5 ± 1.5 kg CH4 h-1 and were significantly affected by factors including the type of feedstock and composting technology, treated feedstock mass, operational patterns and season. The results indicate that the highest methane emission factors were obtained at the combined anaerobic digestion and open windrow composting plant (4.51-5.21 kg CH4 Mg-1 wet garden/park waste (GPW) and food waste), followed by open windrow plants co-composting GPW, sewage sludge and straw (3.49-3.76 kg CH4 Mg-1 wet feedstock). The lowest methane emission factors were found at open windrow composting plants treating GPW (1.56-3.24 kg CH4 Mg-1 wet feedstock). Emissions tended to be higher when measurements were performed during working hours, in comparison to when they were measured after the plant closed for the day. At one plant, emissions were measured monthly over one year, and emissions were about 50% higher in spring and summer in comparison to autumn and winter.

The Danish national effort to minimise methane emissions from biogas plants.

In total, 69 biogas plants representing 59 % of Danish biogas production participated in a national effort to reduce methane (CH4) emission. Measurements in terms of total plant CH4 emissions, quantification of emissions from point sources, leak surveys and conceptual design plans to mitigate emissions were performed. Plant-level CH4 emission rates varied between 1.3 and 81.2 kg CH4 h-1, and CH4 losses expressed in percentages of production varied between 0.3 and 40.6 %. Agricultural plants generally had lower CH4 loss rates compared to wastewater treatment plants. Biogas plants with a smaller gas production emitted a larger fraction of their production compared to larger plants, which was partly explained by the absence of gas collection from digestate storage tanks at smaller plants. A very commonly observed source of emission was pressure relief valves, where this source of leakage was observed at 53 % of the plants. A national emission factor (sum of CH4 emissions/sum of CH4 productions) was determined at 2.5 % for the Danish biogas production, whereof it was 2.1 % for agricultural biogas production and 6.7 % for biogas production at wastewater treatment plants. Measurements of total CH4 emissions at six plants performed before and after implementation of mitigating actions showed that emissions were reduced by 46 % by carrying out relatively minor technical fixes and adjustments. An economic evaluation showed that, in some cases, mitigating actions could be economically beneficial for the biogas plant (positive net present value over a 10 year time frame), due to an increase in revenue.

The effect of barometric pressure changes on the performance of a passive biocover system, Skellingsted landfill, Denmark.

This study investigated the performance of a passive biocover system at a Danish landfill. The overall methane oxidation efficiency of the system was assessed by comparing annual whole-site methane emissions before and after biocover installation. Annual whole-site methane emission predictions were calculated based on empirical models developed by a discrete number of tracer gas dispersion measurements. Moreover, a series of field campaigns and continuous flux measurements was carried out to evaluate the functionality of an individual biowindow. The results indicated that biocover system performance highly depended on barometric pressure variations. Under decreasing barometric pressure, estimated efficiency declined to 20%, while under increasing barometric pressure, nearly 100% oxidation was achieved. In-situ measurements on a specific biowindow showed a similar oxidation efficiency pattern in respect to barometric pressure changes despite the difference in spatial representation. Eddy covariance results revealed pronounced seasonal variability in the investigated biowindow, measuring higher methane fluxes during the cold period compared to the warm period. Results from the in-situ campaigns confirmed this finding, reporting a threefold increase in the biowindow's methane oxidation capacity from April to May. The annual average oxidation efficiency of the system was estimated to range between 51% and 65%, taking into consideration the impact of changes in barometric pressure and seasonal variability. This indicated an annual reduction in landfill's methane emissions between 24 and 35 tonnes. This study revealed the challenge facing current approaches in documenting accurately the performance of a passive biocover system, due to the short-term variability of oxidation efficiency, which is influenced by barometric pressure changes.

Annual upscaling of methane emission field measurements from two Danish landfills, using empirical emission models.

An empirical model was developed and employed to estimate annual methane (CH4) emissions from two Danish landfills (Skellingsted and AV Miljø). The overall aim was to provide accurate annual CH4 emission estimates based on discrete emission field measurements and to address temporal variability caused by the impact of barometric pressure. Four non-linear regression models were developed, corresponding to the two landfills as well as to the western and eastern waste sections of AV Miljø. A comparison of model predictions with on-site eddy covariance fluxes showed that the models can accurately predict short-term emission variability. Predicted annual CH4 emissions for the Skellingsted and AV Miljø landfills were 69 ± 4 and 80 ± 4 tonnes, respectively, whereas for the western and eastern sections of the AV Miljø landfill, emissions were estimated at 63 ± 3 and 19 ± 1 tonnes, respectively. The results demonstrate that even though maximum emissions from Skellingsted were approximately threefold compared to AV Miljø, annual predicted CH4 emissions for Skellingsted were lower. This was because during the most frequently occurring pressure change events, emission rates were higher at AV Miljø in comparison to Skellingsted. An optimised sampling strategy was proposed, targeting the determination of an empirical emission model though the effective use of discrete field measurements. Analysis of annual emission estimates, based on the number of the tracer dispersion method (TDM) measurements, showed that both the number as well as the distribution of performed TDM measurements across the range of expected dP/dt influence the uncertainty.

Methane emissions from Icelandic landfills - A comparison between measured and modelled emissions.

This study compares methane (CH4) emissions from five Icelandic landfills, quantified using tracer gas dispersion to modelled emission rates using the IPCC FOD model. The average CH4 emission rates measured from the investigated landfills were 475.4 kg CH4 h-1 (Álfsnes landfill), 32.5 kg CH4 h-1 (Fíflholt), 40.8 kg CH4 h-1 (Gufunes), 9.8 kg CH4 h-1 (Kirkjuferjuhjáleiga) and 78.4 kg CH4 h-1 (Stekkjarvík). At three of the landfills (Álfsnes, Fíflholt and Kirkjuferjuhjáleiga), the modelled emission was higher than the measured emission by factors ranging from 1.1 to 4.8, neglecting any CH4 oxidation in the cover soils. Even though CH4 oxidation might play a role at some of the investigated landfills, and thus reduce the gap between modelled and measured emissions, it is likely that the model overestimated CH4 generation due to uncertainties in input model parameters. Assuming that the measured emissions at the five landfills are representative of all the waste disposed in Iceland from 2007 to 2016, the measured emission should be extrapolated to 817 kg CH4 h-1, which is relatively close to the modelled national emission of 936 kg CH4 h-1 in 2017. This study showed that the application of the IPCC FOD model at national level is appropriate for estimating landfill CH4 emissions in Iceland. CH4 emissions from landfills in Iceland can be reduced by expanding or implementing gas collection or biocover systems for optimised microbial oxidation.

Evaluation of the toxic effects of ammonia dispersion: consequence analysis of ammonia leakage in an industrial slaughterhouse.

Ammonia is a toxic compound and has many toxic effects on humans and the environment. This study was designed to model the consequences of ammonia leakage in an industrial slaughterhouse. Given the potential hazard of ammonia, only the toxic dimension of this gas was evaluated. The scenarios were evaluated in the worst possible condition and in the case of the complete rupture. Findings showed that in case of a catastrophic rupture scenario in reservoir 1 in the first and second 6 months of the year, the distances of 920.37 and 569.38 m from the reservoir in the wind direction were at Emergency Response Planning Guidelines, level 3 (ERPG3), respectively. In reservoir 2, in the first and second 6 months of the year, the distances of 699.58 and 384.86 m from the reservoir were at the ERPG3 level, respectively. In reservoir 3, in the first and second 6 months of the year, the distances of 203.48 and 748.28 m from the reservoir were at the ERPG3 level, respectively. Examination of the probit values showed that in reservoirs 1 and 2, the probit values were more than 4.28 up to 100 m from the reservoirs, and in the reservoir 3, the mortality rates were lower. The findings revealed that the catastrophic rupture of ammonia reservoirs in the studied slaughterhouse and the release of ammonia could lead to the fatality of large numbers of people in ERPG2 and ERPG3 areas. Therefore, it is necessary to take control measures to reduce the vulnerability against such accidents.

Closing the methane mass balance for an old closed Danish landfill.

In this study, a methane (CH4) mass balance was established for Hedeland landfill. CH4 generation rates were modelled using a multiphase first-order decay model (The Afvalzorg model) and determined at between 57 and 79 kg h-1. The CH4 emission rate was quantified at between 2 and 14 kg h-1, using the tracer gas dispersion method and the CH4 gas recovery efficiency was between 8 and 21%. At three places along the perimeter of the landfill, gas remediation systems have been installed to protect the residential houses from any risk of migrating landfill gas. About 0.76 kg h-1 of CH4 was extracted from these three remediation systems. Using a carbon mass balance for the lateral migrating landfill gas showed a fractional oxidation of about 78%, which corresponded to a CH4 flux of 3.5 kg h-1 from the three remediation systems, including the oxidised CH4. The total lateral CH4 flux (un-oxidised) from the total landfill perimeter was estimated at between 6.9 and 10.4 kg h-1. CH4 oxidation efficiency in the landfill cover soil, determined from stable carbon isotope analyses, was found to be between 12% and 92%. This resulted in an average CH4 oxidation rate of 32 kg h-1, using an average CH4 emission rate of 8 kg h-1. CH4 surface screenings and surface flux measurements supported the hypothesis that oxidation efficiency was in the higher range and that oxidation could close the CH4 mass balance.

Total methane emission rates and losses from 23 biogas plants.

Methane losses from biogas plants are problematic, since they contribute to global warming and thus reduce the environmental benefits of biogas production. Total losses of methane from 23 biogas plants were measured by applying a tracer gas dispersion method to assess the magnitude of these emissions. The investigated biogas plants varied in terms of size, substrates used and biogas utilisation. Methane emission rates varied between 2.3 and 33.5 kg CH4 h-1, and losses expressed in percentages of production varied between 0.4 and 14.9%. The average emission rate was 10.4 kg CH4 h-1, and the average loss was 4.6%. Methane losses from the larger biogas plants were generally lower compared to those from the smaller facilities. In general, methane losses were higher from wastewater treatment biogas plants (7.5% in average) in comparison to agricultural biogas plants (2.4% in average). In essence, methane loss may constitute the largest negative environmental impact on the carbon footprint of biogas production.

Development and implementation of a screening method to categorise the greenhouse gas mitigation potential of 91 landfills.

A cost-effective screening method for assessing methane emissions was developed and employed to categorise 91 older Danish landfills into three categories defined by the magnitude of their emissions. The overall aim was to assess whether these landfills were relevant or irrelevant with respect to methane emission mitigation through the construction of biocovers. The method was based on downwind methane concentration measurements, using a van-mounted cavity ring-down spectrometer combined with inverse dispersion modelling to estimate whole-site methane emission rates. This method was found to be less accurate than the more labour-intensive tracer gas dispersion method, and therefore cannot be recommended if a high degree of accuracy is required. However, it is useful if a less accurate examination is sufficient. A sensitivity analysis showed the dispersion model used to be highly sensitive to variations in input parameters. Of the 91 landfills in the survey, 25 were found to be relevant for biocover construction when the methane emission threshold was set at 2 kg CH4 h-1.

Methodologies for measuring fugitive methane emissions from landfills - A review.

Fugitive methane (CH4) emissions from landfills are significant global sources of greenhouse gases emitted into the atmosphere; thus, reducing them would be a beneficial way of overall greenhouse gas emissions mitigation. In Europe, landfill owners have to report their annual CH4 emissions, so direct measurements are therefore important for (1) evaluating and improving currently applied CH4 emission models, (2) reporting annual CH4 emissions and (3) quantifying CH4 mitigation initiatives. This paper aims at providing an overview of currently available methodologies used to measure fugitive CH4 emissions escaping from landfills. The measurement methodologies are described briefly, and the advantages and limitations of the different techniques are discussed with reference to published literature on the subject. Examples are given of individual published studies using different methodologies and studies comparing three or more methodologies. This review suggests that accurate, whole-site CH4 emission quantifications are best done using methods measuring downwind of the landfill, such as tracer gas dispersion and differential absorption LiDAR (DIAL). Combining aerial CH4 concentration measurements from aircraft or unmanned aerial vehicles with wind field measurements offers a great future potential for improved and cost-efficient integrated landfill CH4 emission quantification. However, these methods are difficult to apply for longer time periods, so in order to measure temporal CH4 emission changes, e.g. due to the effect of changes in atmospheric conditions (pressure, wind and precipitation), a measurement method that is able to measure continuously is required. Such a method could be eddy covariance or static mass balance, although these procedures are challenged by topography and inhomogeneous spatial emission patterns, and as such they can underestimate emissions significantly. Surface flux chambers have been used widely, but they are likely to underestimate emission rates, due to the heterogeneous nature of most landfill covers resulting in sporadic and localised CH4 emission hotspots being the dominant emission routes. Furthermore, emissions from wells, vents, etc. are not captured by surface flux chambers. The significance of any underestimation depends highly on the configuration of individual landfills, their size and emission patterns.

Determination of gas recovery efficiency at two Danish landfills by performing downwind methane measurements and stable carbon isotopic analysis.

In this study, the total methane (CH4) generation rate and gas recovery efficiency at two Danish landfills were determined by field measurements. The landfills are located close to each other and are connected to the same gas collection system. The tracer gas dispersion method was used for quantification of CH4 emissions from the landfills, while the CH4 oxidation efficiency in the landfill cover layers was determined by stable carbon isotopic technique. The total CH4 generation rate was estimated by a first-order decay model (Afvalzorg) and was compared with the total CH4 generation rate determined by field measurements. CH4 emissions from the two landfills combined ranged from 29.1 to 49.6 kg CH4/h. The CH4 oxidation efficiency was 6-37%, with an average of 18% corresponding to an average CH4 oxidation rate of 8.1 kg CH4/h. The calculated gas recovery efficiency was 59-76%, indicating a high potential for optimization of the gas collection system. Higher gas recovery efficiencies (73-76%) were observed after the commencement of gas extraction from a new section of one of the landfills. A good agreement was observed between the average total CH4 generation rates determined by field measurements (147 kg CH4/h) and those estimated by the Afvalzorg model (154 kg CH4/h).

Comparison of Machine Learning Models for Hazardous Gas Dispersion Prediction in Field Cases.

Dispersion prediction plays a significant role in the management and emergency response to hazardous gas emissions and accidental leaks. Compared with conventional atmospheric dispersion models, machine leaning (ML) models have both high accuracy and efficiency in terms of prediction, especially in field cases. However, selection of model type and the inputs of the ML model are still essential problems. To address this issue, two ML models (i.e., the back propagation (BP) network and support vector regression (SVR) with different input selections (i.e., original monitoring parameters and integrated Gaussian parameters) are proposed in this paper. To compare the performances of presented ML models in field cases, these models are evaluated using the Prairie Grass and Indianapolis field data sets. The influence of the training set scale on the performances of ML models is analyzed as well. Results demonstrate that the integrated Gaussian parameters indeed improve the prediction accuracy in the Prairie Grass case. However, they do not make much difference in the Indianapolis case due to their inadaptability to the complex terrain conditions. In addition, it can be summarized that the SVR shows better generalization ability with relatively small training sets, but tends to under-fit the training data. In contrast, the BP network has a stronger fitting ability, but sometimes suffers from an over-fitting problem. As a result, the model and input selection presented in this paper will be of great help to environmental and public health protection in real applications.

The real-time estimation of hazardous gas dispersion by the integration of gas detectors, neural network and gas dispersion models.

Release of hazardous materials in chemical industries is a major threat to surrounding areas. Current gas dispersion models like PHAST and FLACS, use release velocity, release elevation, meteorological parameters, and other related information as model input. In general, such information is not always available during an on-going accident. In this paper, we develop a fast prediction approach which could bypass the input parameters that are difficult to obtain and predict the released gas concentration at certain off-site location using parameters that could be obtained easily. The new approach is an integration of gas detectors, artificial neural network (ANN) and one of the aforementioned gas dispersion models. PHAST is applied to simulate numbers of release scenarios and the results containing the spatial and temporal distributions of released gas concentration are prepared as input and target data samples for training the neural network. The approach was applied to a case study involving a hypothetical chlorine release with varying release rates and atmospheric conditions. The results of the approach that are concentration and dispersion time profiles in the environmental sensitive locations were validated against PHAST. The validation shows highly correlations with PHAST and convincingly demonstrates the effectiveness of the proposed approach.

Effect of konjac glucomannan addition on aroma release in gels containing potato starch.

The present study aimed to measure the retention of aroma compounds (ethyl acetate, ethyl hexanoate and carvacrol) in dispersions based on konjac glucomannan and/or potato starch, and to highlight the influence of konjac glucomannan on the mechanisms involved in aroma retention. Publications on the effect of konjac glucomannan on aroma release are scarce. Konjac glucomannan is a polysaccharide used as a food additive for its viscous and emulsifying properties. Retention of aroma compounds in dispersions was calculated from partition coefficients which were measured using the phase ratio variation method. This method, consisting of analyses of the headspace at equilibrium, enables the determination of the partition coefficient of volatile compounds in a gas/liquid system without external or internal calibration. The three aroma compounds chosen for this study behave differently toward amylose. Prior to the release study, the complexing behavior of carvacrol with starch, hitherto unknown, was investigated by X-ray diffraction: V6III amylose complexes were formed with carvacrol. Our results showed no specific interaction between ethyl hexanoate and potato starch or konjac glucomannan. Ethyl acetate retention seemed to be due to trapping in the complex network of polysaccharides and to the density of this network. Retention of carvacrol was influenced by the nature of polysaccharides present in the dispersion, and was mainly governed by specific interaction with starch. Additionally, the addition of konjac glucomannan to potato starch dispersions decreased the retention of volatile compounds complexing starch, but had little effect on the retention of the other aroma compounds.