MODELLING OF UASB REACTOR

Abstract

Anaerobic digestion is the biological stabilization of organic substrates under anaerobic conditions. During the process, organic wastes are stabilized and biomass is produced with a substantial net reduction in the solid mass. The chemical oxygen demand (COD) reduction is mainly due to methane production. The anaerobic digestion has received a lot of significance in the treatment of wastewaters since the introduction of the concept of sludge retention for a longer time in the early sixties. Different types of anaerobic digestion units have been developed and among these, Upflow Anaerobic Sludge Blanket (UASB) reactor system has received wide recognition. In the last two and half decades, the upflow anaerobic sludge blanket (UASB) reactor has been recognized as an efficient and energy saving anaerobic wastewater treatment system. Since its development in 1970s by Lettinga and co-workers in Netherlands, it has been used for the treatment of a variety of liquid wastes of industries such as sugar, alcohol and distillery, food processing, dairy, petrochemicals, pharmaceuticals, pulp and paper, and of slaughterhouse and domestic sewage. Due to the requirement of high cost involved in the setting of a UASB reactor and the problems associated with its operation, it is necessary to study the various factors affecting its design and performance. Inspite of numerous applications of UASB reactors [Lettinga et al. (1980), Cail and Barford (1985), and Fang ct al. (1995a)] in the treatment of organic liquid wastes successfully, operational difficulties still persist and reactor failures have been encountered in several cases. Review of literature indicates that the start-up phase in UASB reactors can be as high as nearly one year. Thus, one has to consider the start-up phase as one of the important issues to be addressed in studies on UASB reactor. Main emphasis in the present research work is on the following aspects, viz., development of start-up strategies to enhance granulation or expedite the granulation phase, settling characteristics of the granules, modelling of flow distribution parameters in a UASB reactor, and simulation of a variety of experiments on UASB reactor with due emphasis to the simplified procedure of UASB reactor performance simulation. On the basis of the analysis of available experimental data and existing status on modelling of UASB reactor, the following important objectives were formulated: (1) To arrive at unified definition of the granulation time for study of the influence of organic loading rate (OLR), sludge loading rate (SLR) and nutrients dosing, and identification of favourable ranges of these variables for expediting the granulation/ start-up; (2) To study the settling characteristics of granules and to provide flow rate regulation for speedy granulation; (3) To study distribution of flow at the bed, blanket and settler levels using dimensional approach and to evaluate the efficacy of this approach or methodology in real practice, and (4) Simulation of UASB reactor performance data on a variety of wastes including, whey permeate wastewater, cornstarch wastewater, and domestic wastewaters, in which the breakdown and stoichiometric relationships are not adequately known from the available literature. To improve the UASB reactor start-up phase, the influence of several variables on granulation time was studied for variety of wastes. For this purpose, the available experimental data were collected and/ or processed to obtain their representative values during granulation. A multiple regression analysis between the granulation time and some of the important variables was also attempted but in view of its limited success, the isolated variation of granulation time with variables like organic loading rate, sludge loading rate and seed/ biomass concentration were studied to extract the following conclusions. (i) Organic loading rate (OLR) regulation as per the existing criteria did not correspond to optimum granulation in case of different wastes including volatile fatty acids, sugar molasses, pulp and paper mill, food processing wastewaters, sugar/ glucose wastes, etc. New organic loading rate limits have been obtained to expedite granulation in case of several wastes. The proposed criteria is based on a set of 25 experimental investigations in comparison to the few experiments based existing criteria of OLR regulation. (ii) Based on the present study [Singh et al. (1998a)], it is suggested that the higher limit of sludge loading rate at the time of granulation can be as low as 0.2 g COD/g VSS.d (food processing wastes) and as high as 0.86 g COD/g VSS.d (volatile fatty acids). The observed biomass concentration at G, could be even as low as 6.0 g VSS/1 and as high as 23.0 g VSS/1 depending upon the waste types and seed materials used in the beginning of the process. (iii) The initial seed concentration can be as low as 6.0 gVSS/1 (sugar/ glucose wastewater), which is lower than the value suggested in the literature. Similarly, the higher initial seed concentration observed in this study was 25.00 gVSS/1 (volatile fatty acids), which is on the higher side to the suggested values. This difference could be attributed to the nature and type of seeds used, their activity and characteristics of the waste. (iv) Nutrients as well as tracemetals dosing in terms of per gm COD of wastes have been given [Singh et al. (1998b)]. For sodium, potassium, aluminium and yeast extracts, the range of uncertainty in which the doses are to be prescribed is very less, For nickel, the range is widest. As adequate emphasis has not been given to nutrients dosing in the literature, useful figures are developed to guide the user to select nutrients and tracemetals doses. Higher nutrient doses do not necessarily correspond to the minimum granulation time, and wherever available, such nutrient doses are also provided in these figures. (v) A systematic start-up strategy has been evolved on the basis of the results obtained and using the experiences of other researchers, which has a practical utility in solving the start-up associated problems in UASB reactors. Settling characteristics of granules have been studied by treating granules to be spherical in shape. A composite structure of the granule defined in terms of a core of heavier density and a flocculent sheath of lighter density around the core has been assumed. As the granules may escape the UASB reactor from the settler zone, the conditions prevailing in the settler in terms of the low concentration of the granules have been used to develop the settling characteristics. Considering the ratio of granule diameter to the reactor diameter, the role of wall effect has also been ignored. Nomographs have been prepared for quick evaluation of the settling velocity of granules for core diameters range of 0.5 to 2.0 mm, density of core material to range from 1010 to 1080 kg/ m , and density of flocculent material to range from 1001 to 1005 kg/ m . The proposed composite structured model of granule is found to reproduce some of the available results of settling velocity. Knowledge of granule diameters during operation coupled with the developed nomographs of the settling velocity is likely to help the flow regulation in UASB reactors. An issue very closely related to the settling characteristics of granules is the variation of the ratio of organic loading rate at any time within the granulation period to that applied at the beginning of the process operation. Based on the observations of organic load variation of certain wastes in which the granulation time was of the order of 4 to 6 weeks, the exponential variation of organic loading rate has been proposed. This variation of organic loading rate has to satisfy that the upflow velocity remains below settling velocity. There is no clear-cut information regarding the granule size variation with respect to time in an UASB reactor and this aspect still requires further investigations. The applied input load to the UASB reactor is seldom evenly distributed and it experiences short-circuiting. The short-circuiting portion of inflow can either partially re enter the blanket zone or it can enter the settler zone. Furthermore, there is a back mix flow from the blanket to bed due to uprising gas bubbles. Short-circuiting of the inflowas well as the backmixing flow has been found related to the biomass quantity/ concentration in the reactor, reactor dimensions, and fluid flow characteristics. The flow distribution parameters arc found related to certain dimcnsionless numbers. Interesting variation of the flow distribution parameters with dimcnsionless groups has been obtained. The present study has offered a better approach for modelling of short ly circuiting or backmix flows in terms of dimensionless variables. Also, some of the available relationships have been found questionable. Flow pattern in UASB reactors is complex, and may depend upon its geometry, properties of granules, and operating parameters. In recent years, several studies have focussed on UASB reactor modelling with emphasis on the use of comprehensive models based on the UASB reactor approximation as a series of CSTR and plug flow reactor combinations [Van Der Meer and Heertjes (1983), Bolle et al. (1986b and 1986c), and Wu and Hickey (1997)]. However, in several other studies [Bhatia et al. (1985b), Costello et al. (1991a), and Hwang et al. (1992)], UASB reactor performance has been found close to a CSTR. In the literature, not enough evidence is available to reflect the limitations of single CSTR models of UASB reactor. Also the role of bypassing stream in the reactor has not been given reasonable attention in many studies including Hwang et al. (1992). For this reason, the present study has looked into simulations of laboratory and industrial scale UASB reactors' performance using a CSTR model with or without bypassing stream. The various wastes considered for simulation are- whey permeate wastewater, cornstarch wastewater, etc. For simplicity, the model assumes Monod kinetics expression [Narasiah (1983), Tien (1994) and Kus and Wiesmann (1995)] for microbial growth. The values of flow distribution parameters obtained in these simulations are found related to the biomass concentration within the reactor, and this trend is also supported from the tracer test results of UASB field scale plant of Van Der Meer (1979). Based on simulation studies, it appears that idealization of UASB reactor as single CSTR may not be always a better option. Therefore, it is proposed that more complex flow pattern models [Bolle et al. (1986c)] should be used for simulating a real UASB reactor. But for such simulations, a lot of planning for acquiring data is very much necessary. Also, it is recommended that the knowledge of correct kinetic parameters, a periodical tracer test information, and regular monitoring of biomass concentration within and in the effluent of reactor is must for simulating the performance of UASB reactor.