OPTIMIZATION OF MUNICIPAL WASTEWATER TREATMENT SYSTEM WITH ACTIVATED SLUDGE PROCESS

Abstract

The research embodied in this thesis deals with the development of a rationale for the selection of minimum cost design of municipal wastewater treatment plants generally employed in urban sanitation and water pollution control programmes. The conventional method of designing a wastewater treatment system consists in sizing different units indivi dually and then placing these units together to form a complete system. Such a method seldom yields an optimal design. Systems approach in engineering design, involving optimization of total system function with recourse to known optimization techniques, has now come to stay as a useful tool for providing optimal design of the treatment systems. Synthesis of optimal design of wastewater treatment systems in general and activated sludge system in particular has been reported extensively during the last several years. However this research addresses itself to the working and installation of these plants in non-tropical countries and accordingly the studies relevant to Indian conditions at this point of time when the country is poised for extensive water pollution control programmes could provide substantial savings in Government and international investments for meeting the objectives of International Drinking Water ii Supply and Sanitation Decade. The thesis presents algorithms for optimal design for activated sludge plants treating municipal wastewaters to meet the stipulations of Water Pollution Control Authori ties in India. Specifically the system investigated includes a liquor stream treating screened and degritted municipal sewage led to an activated sludge system, and a sludge stream employing a thickener followed by anaerobic digester and subsequent dewatering of digested sludge. The system is subjected to influent variability of flow and organic load. The basic objectives of the study presented in the thesis are delineated hereunder: (i) to tackle influent variability with a view to minimising system disorders and plant effluent variability, and include its consideration in the total system modelj (ii) to review design criteria and existing mathematical models for the design of various treatment units incorporated in the system and to generate a total system model to suit specific environmental conditions in Indiaj (iii) to develop atotal cost model for the system at the prevailing rates of man, material and power in India, and Civ] to develop algorithms for minimum cost design whilsatisfying the various d sign constraints- The treatment system is conventionally designed on the Ill basis of mean value inputs and steady state conditions- Inherent variability in the flow and load characteristics of the sewage received at plant, however, results in frequent operational disorders and consequent fluctuations in the plant effluent characteristics. The need for attenuat ing input variability in organic load before the sewage is led to the biological treatment system, therefore, can not be overemphasised. Introduction of an equalisation basin at the head of the plant has been advocated for controlling effluent variability. The present study examines the efficacy of the equaliza tion basin in the overall system design with recourse to the data collected at Okhla sewage treatment plant at Delhi. The equalisation basins are designed either by the U.S. Environ mental Protection Agency (EPA) method of balancing storage considering influent variability and acceptable variability in the biological reactor, or the equalisat ion of time dependent wasteloads based upon power spectrum estimates as advocated by Novotony et al. Acomparative study of these methods brings out that the Novotony»s approach yields higher variance removal (of 30D5 load) than that provided by the EPA approach. The difference is attributed to the fact that while the former considers both decay and dilution processes in equalization, -the latter considers dilution only. An additional advantage of Novotony«s approach is in its IV complete depiction of input characteristics in the form of power spectrum and amplitude gain factors, which could also be utilised in arriving at effluent variability of the total system. A detailed review of the basic kinetic and solidliquid separation equations along with the empirical relationships presently used in design have also been presented in this work to arrive at functional relationships userai in the design of each Unit of trie treatment system. The formulation of functional model for the treatment system useful in the Indian context warranted consideration of relevant environment factors. Accordingly a comparative study has been conducted of the data reported in North India for primary settling unit with the CIRIA model. This study confirmed author's contention and resulted in a functional model for "the primary settler to be incorporated in -the total system model. Ibr tlie design of activated sludge system, Gaudy's model has been preferred over other models (as it is easier to control the biological reactor) and, therefore the mass balance has been written around the reactor, wherein recvclo ratio (a) and returned sludge concentration (Xr) have been directly considered- Energy requirements, for maintaining completely mixed regime and limiting D.O. for ensuring proper settling characteristics of the mixed liquor, have been calculated as per the specifications of surface aerators being manufactured in India. Accordingly a relationship between power requirement and MLSS concentration has been arrived at and utilised in the system model. A rational design of secondary settling unit essentially requires data from batch settling tests. Accordingly abatch settling study has been done to derive relationships for different design MLSS values by taking samples from aeration tank of Okhla plant. A relationship between limiting solids flux and underflow concentration UJ has been established and used in the system model- In the sludge stream, the thickener has been designed on average solids loading considerations. This unit is foll owed by an aerobic digestion unit in the treatment train. Both the conventional and high rate digestion systems have been considered separately to assess their economic viability in the total system design. Heating during winter months in the digester has been taken into account. The high rate digestion system has been designed on the basis of mass balance consideration around the reactor. For designing drying beds, Arceivala's model, which provides for direct consideration of rainfall and evaporation rates, has been included in the system model for rainfall and evaporation rate data from Delhi region. The cost function of a treatment system comprises of vi the cost of manpower, energy, equipments and raw materials. The contribution of these items in the total cost differs from country to country e.g. at 1977 index, percentage allocations for personnel and power in sewage treatment plants of 68 to 90 mid size range have been reported as 48 and 27 respectively in USA, as against 28 to 35 and 55 to 65 respectively in India. The success of an excercise in optimization depends upon the development of a rational cost model applicable to specific situation under consideration. Hence, a cost model has been evolved for Indian conditions at 1981-82 market index- The cost data for the equipment as well as operation and maintenance have been collected from Indian manufacturers and records of existing activated sludge plants in India. The energy costs have been calculated at prevailing rates. The civil costs have been assigned by working out structural designs as per Indian standard code of practice at the Central Public Works Department (Delhi region) rates-The need for adopting this methodology for civil works costing arose becausa the quotations gathered from various consultants stationed at various parts of the country differed significantly. The capital and operating cost data for each unit have been utilised to generate a total system cost function in the form of Total Annual Cash Flow (TACF) which incorporates important economic parameters like interest and depreciation rates. vi: Optimal design of the treatment plant is,therefore,defined as that design which yields minimum TACF value of the system within stipulated constraints- The constraints incorporated are the limiting values of design parameters viz- maximum allowable BOD5 value in the plant effluent any time with 98 percent confidence limit, detention time in primary settler, food to micro-organisms ratio (P/M ratio) and mean cell residence time (MCRT) value in aeration tank, overflow rate in secondary settler, prethickening of sludge before digestion, and volatile solids loading in digester- With a view to avoiding waste of computer time in an optimization study, it is often desirable to carry out a sensitivity analysis to identify significant design variables alongwith the range of variation- This has been accomplished by carrying cut acomputer aided parametric study, on the total system for the data collected from Okhla plant on DEC SYSTEM-20. The important design variables identified for the study are the detention time (DTEB) in the equalisa tion basin, overflow rate in primary settler (PSOR), mixed liquor suspended solids concentration (MLSS) in aeration tank, underflow concentration C«r) in secondary settler end solids retention time (SRT) in anaerobic digester. Results of parametric study reveal that the inclusion of high rate digester makes the total System more economical than the system with aconventional digester- Further, DTEB, PSOR and MLSS are more significant and viii sensitive design variables than Xr and SRT with respect to TACF value of the system. Asignificant result obtained in this work is the capability of the activated sludge process and primary settler in providing adequate equalisation to satisfy the effluent variability criteria as against the provision of a separate equalisation basin in the municipal wastewater treatment system. Another important result of the study is that the least cost design corresponds to the lowest PSOR values in primary settler (or it is less expensive to divert higher amounts of settled solids to the sludge stream rather than to the liquid stream).The results of parametric study have brought out the convex character of the objective function with respect to all design variables considered in this study. The nonlinear character of objective function (TACF) warrants application of nonlinear optimization technique for the identification of minimum cost solution. Two direct search nonlinear unconstrained optimization techniques viz- Univariate method and Powell's method have been employed in this research. Use of Powell's method has been reported in the literature while the former technique has been used in this work to study the relative rates of convergence of the two methods. Both algorithms have been found to be suitable and satisfy the convergence criteria quickly while arriving at global optimal designs because of the convex nature of the TACF function. The optimal designs have been compared with convent ional design (by using average values of design variables from the recommended ranges in the existing practice^ Optimal design delineated in parametric study is 9-9 percent cheaper while the ones synthesized through Powell's and Univariate methods are 18-83 and 19-76 percent respectively compared to the conventional design. The CPU time on DEC-20 SYSTEM was 20-46 seconds in parametric study, 3-21 seconds in Powell's method and 6-02 seconds in Univariate method. Thus the Powell's method has been found to be better than the Univariate method in respect of the convergence although the Univariate method provides marginally cheaper design. The thesis, thus, presents a methodology for synthesising optimal designs for municipal wastewater activated sludge systems under time dependent input values in Delhi region. The algorithms could be utilised in the design of plants elsewhere in India with appropriate modifications in input parameters and cost model.