外文原文-选煤厂优化现有优化方法的修正.pdf

Coal preparation plant optimization A critical review of the existing methods V Gupta 1 M K Mohanty Department of Mining and Mineral Resources Engineering Southern Illinois University at Carbondale Carbondale Illinois 62901 6603 United States Received 19 July 2005 received in revised form 13 November 2005 accepted 15 November 2005 Available online 4 January 2006 Abstract A coal preparation plant typically operates with multiple cleaning circuits to clean individual size fractions of run of mine coal Coal preparation plants are traditionally optimized using the equalization of incremental product quality approach Individual cleaning circuits are operated at the same specific incremental product quality so that the targeted overall plant product quality is achieved Over the years it has been well established that equal incremental product quality approach maximizes plant yield for a given product quality constraint However while dealing with multiple quality constraints the incremental quality approach may not provide a complete solution to the optimization problem It may be intuitive to realize that the dirtiest particle s in a coal product with respect to ash content may not be the same particle s with respect to sulfur content Therefore with increasing number of product quality constraints which may include but not limited to limiting ash sulfur and trace element contents the plant has to be optimized based on each incremental product quality Understandably the operating points selected for each circuit to maximize plant yield based on incremental ash content may not be suitable for obtaining maximum plant yield based on incremental sulfur content These limitations of the equalization of incremental product quality approach to satisfy multiple product quality constraints have been reviewed in detail in this publication with an example of ash and sulfur data collected from an operating coal preparation plant D 2005 Elsevier B V All rights reserved Keywords coal plant optimization incremental product quality coal washability 1 Introduction Based on the size consist of the Run of mine ROM coal a preparation plant utilizes three or four individual circuits to clean the entire ROM coal For example coal coarser than 12 5 mm may be cleaned in a heavy medium vessel circuit 12 5 1 mm in a heavy medium cyclone circuit 1 mm 150 Am in a spiral circuit and minus 150 Am size coal in a flotation circuit Typically product quality measure such as ash content from each circuit is maintained at nearly the same level as the target ash content for the overall plant In other words if a plant contract requires product specifications of 8 ash the operating conditions in the individual circuits are adjusted so that the ash contents of the individual circuit products are approximately 8 Although this approach of producing equal average product quality 0301 7516 see front matter D 2005 Elsevier B V All rights reserved doi 10 1016 j minpro 2005 11 006 Corresponding author Tel 1 618 453 7910 fax 1 618 453 7455 E mail address mohanty engr siu edu M K Mohanty 1 Tel 1 618 453 7910 fax 1 618 453 7455 Int J Miner Process 79 2006 9 17 from each circuit provides a simplistic solution to sa tisfy contract specifications it does not guarantee the maximum possible plant yield Incremental product quality concept is commonly used to maximize plant yield for a given quality con straint By definition incremental product quality refers to the quality of the dirtiest particle s present in any coal product whereas the average product quality refers to the overall quality of the composite coal product Numerous studies have been conducted in the past to develop suitable procedures for maximizing overall plant yield while satisfying a desired average product quality Sarkar et al 1960 suggested a graphical ap proach for maximization of yield of composite clean coal at a desired ash content It was suggested that cleaning of coarser coal at a higher ash content and finer coal at a relatively lower ash content gives the maximum yield while satisfying the given product ash constraint Walters and Ramani 1976 developed a computer based plant optimization model in which the separating gravity of the small coal is held constant while the separating gravity of the coarse coal is incre mented until the desired product quality is achieved Abott 1982 derived an equation to prove that the optimum conditions for maximum profit from a blend of coal produced by two different cleaning processes occurs when the instantaneous ash incremental ash contents of both clean products are equal Salama 1986 1991 and 1998 and King 1999 developed graphical and numerical techniques to optimize the yield of a plant at a given product quality constraint Graphical methods were based on the Mayer curve M curve to determine the optimum cut points of separa tion which maximized the plant yield at a given product ash content Rayner 1987 also utilized the graphical technique of plotting M curve for yield maximization of a plant at a given ash constraint Romberg 1990 developed the optimization software COALTROL uti lizing the M curve for yield maximization These graphical approaches have limitation since complexity of a coal preparation plant increases with an increase in the number of cleaning circuits Salama 1986 verified mathematically that equalization of incremental ash gives the optimum yield at a given product ash con straint Lyman 1993 developed a computational model based on the incremental ash approach for yield maximization in a coal preparation plant Subse quently Salama and Mikhail 1994 also developed a simulation model for plant yield maximization based on the incremental ash approach Sen et al 1994 derived a mathematical solution utilizing a Lagrangian function for cleaning coal obtained from multiple sources and arrived at the same conclusion which recommended the equalization of incremental product quality to achieve optimum plant yield Luttrell et al 2003 2004 also arrived at the same conclusion that the incremental ash approach gives the maximum yield while satisfying a given ash constraint By showing a direct relationship between the ash content of a coal particle and its density Luttrell et al 2003 2004 claimed that the plant performance could be optimized by operating the plants at the same specific gravity cut point Although from their plot shown in Fig 1 a good correlation between ash content and specific gravity is quite evident the data scatter may very well be resulting due to varying composition of the coal ash which might have been overlooked by the investigators A simple analysis of coal ash may indicate the presence of a variety of minerals including SiO2 Al2O3 Fe2O3 CaO K2O and others Although SiO2 having a specific gravity of 2 65 constitutes more than 50 of the total ash material the concentration of Al2O3 SG 3 9 to 4 1 and Fe2O3 SG 5 0 to 6 0 in coal ash is also quite significant In some coal obtained from Illinois No 6 seam the combined concentration of these two miner als Al2O3and Fe2O3 is as high as 30 PHYLISS data Base 2005 Understandably the presence of dif ferent proportions of these minerals in a coal would allow specific coal or specific size fractions of a coal to have different density even when their ash contents are the same and vice versa Almost all of the past investigators discussed plant optimization with single product quality constraint with the exception of Honaker et al 1997 who looked into multiple product quality constraints al though not simultaneously The equalization of incre 100 80 60 40 20 0 Plus 1 1 4 1 1 4x3 4 3 4x1 4 1 4x28M 28x48M 48x100M 8 0 0 4 5 7 5 0 10 2 Ash Content 1 Specific Gravity 0 30 40 50 60 70 80 9 Fig 1 Relationship between specific gravity and ash content for various size fractions of a run of mine coal Luttrell et al 2003 2004 V Gupta M K Mohanty Int J Miner Process 79 2006 9 1710 mental product quality approach was utilized by Honaker et al to maximize the plant yield values on the basis of incremental ash incremental sulfur and incremental trace element contents in succession However no suitable method was prescribed to max imize the plant yield while simultaneously satisfying multiple quality constraints as indicated in their plots shown in Fig 2 The maximized plant yield values shown in Fig 2 a over a range of product ash contents appear to be the same yield values shown over a range of ash and sulfur contents in Fig 2 b It appears from these plots that the maximized yield values were determined by equalizing incremental ash contents in individual plant circuits followed by the determination of the average ash and sulfur contents for the overall plant product at those operating points In other words the maximized yield values obtained from the equalization of incremental sulfur content of individual circuits was not given any consideration in the plot shown in Fig 2 b The present study investigates the limitations of dealing with multiple product quality constraints dur ing plant optimization in greater details Actual ash and sulfur data obtained from the tests conducted for each unit operations of a four circuit coal preparation plant have been utilized as example to illustrate the limitations of the incremental product quality ap proach Consequently the study recommends the ne cessity of a better optimization method to maximize plant yield especially to satisfy multiple product qual ity constraints 2 Experimental A four circuit plant investigated in this study uses heavy medium to clean all of the coal coarser than 1 mm A heavy medium vessel HMV is utilized to clean the plus 16 mm size fraction along with a heavy medi um cyclone HMC cleaning the 16 1 mm size frac tion of the run of mine coal The 1 mm 150 Am and minus 150 Am size coal fractions are cleaned using banks of 3 start spirals and froth flotation cells respec tively The washability data of feed reporting to all three density based separation circuits i e HMV HMC and Spirals are provided in Table 1 along with the flotation kinetic data for the flotation circuit feed As it is shown the heaviest fraction 2 0 sink in feed is significantly higher for HMC unit than the other two density based separators This indicates that the expected product yields from HMC would be significantly lower than HMV and spiral Characteristic partition curves were fitted to the performance data obtained from at least five actual tests conducted in each cleaning circuit by varying the key operating condition Medium density was var ied in case of HMVand HMC whereas splitter position and froth height were varied for the spirals and flotation cells respectively However apparently due to fluctua tions in the plant feed along with possible sampling errors the flotation tests did not produce any meaning ful data Hence it was decided to simulate the flotation performance based on laboratory flotation kinetic anal ysis of the sample of actual feed slurry reporting to the flotation cells in the plant 3 Results and discussion ThemodifiedLynchequation Eq 1 was successfully fitted to the normalized partition data 90 85 80 75 70 65 60 55 50 Conventional Yield Optimized Yield Advanced Circuit I Advanced Circuit II 456 Clean coal yield Product Ash 78910111213 85 80 75 70 65 Conventional Yield Optimized Yield Advanced Circuit I Advanced Circuit II 5 0 97 6 1 06 7 1 11 7 5 1 18 8 1 12 8 5 1 12 9 1 12 10 1 12 Clean coal yield Multiple Constraint Points Ash Sulfur b a Fig 2 Maximized plant yield values obtained from an optimization model while satisfying a one quality ash constraint and b two quality ash and sulfur constraints Honaker et al 1997 V Gupta M K Mohanty Int J Miner Process 79 2006 9 1711 obtained from the tests conducted for HMV HMC and coal spirals PN float eax 1 eax ea 2 1 where xnormalized specific gravity mean specific grav ity specific gravity of separation afitting constant Although the nature of the model equation remained the same the fitting constant baQ was different for all three density based separators Using these model equa tions characteristic yield ash and yield sulfur data points were generated using the washability data Table 1 of feed coal reporting to each cleaning circuit at very close separation density D50 intervals These data were utilized to calculate the incremental product ash and product sulfur contents corresponding to each separation density using the following equation IGj 1 Yj 1Gj 1 YjGj Yj 1 Yj 2 where Yj Gjyield and grade at jth density cut point or sepa ration density Yj 1 Gj 1yield and grade at the next higher i e j 1 th density cut point IGj 1incremental grade at j 1 th density cut point Pursuing the traditional optimization approach the incremental product quality was equalized from all four cleaning circuits to maximize the plant yield A simple illustration of the approach is provided in Fig 3 a b and c As shown in Fig 3 a the incremental ash content was equalized at an arbitrary value of 22 to produce mass yields of 63 84 42 70 67 05 and 47 42 from HMV HMC spiral and flotation circuits respectively The corresponding average product ash contents of 7 03 7 12 4 71 and 9 97 for the respective cleaning circuits were determined from the yield ash curves for each circuit as shown in Fig 3 b Subse quently the overall plant yield and plant product ash contents of 54 67 and 6 73 were calculated as the weighted average of individual circuit yield and product ash contents If the desired plant product ash content is also 6 73 then the maximum plant yield achievable would be 54 67 The expected plant product sulfur content of 1 15 would be determined as the weighted average of the individual circuit sulfur contents as illustrated in Fig 3 c However if the desired product ash content was different from 6 73 then similar iterations were carried out at lower if the desired product ash is lower than 6 73 or higher if the Table 1 Washability data for the three density based cleaning circuits and the flotation kinetics data for the flotation cleaning circuit of a 4 circuit plant studied during this investigation Heavy media vesselHeavy media cyclone Sp gr Weight Ash Sulfur Sp gr Weight Ash Sulfur 1 15 1 251 482 750 801 15 1 2524 204 201 00 1 25 1 332 764 480 901 25 1 311 358 041 23 1 3 1 427 018 751 021 3 1 44 9013 172 30 1 4 1 52 3215 223 411 4 1 50 9619 513 67 1 5 1 6250 5630 214 131 5 1 6251 8620 722 80 1 625 1 91 5639 473 031 625 2 02 7958 142 87 1 9 2 834 3293 620 622 0 2 853 9584 280 40 Total100 037 140 94Total100 050 240 88 SpiralFlotation Sp gr Weight Ash Sulfur Time s Weight Ash Sulfur 1 15 1 345 212 160 8700 000 000 00 1 3 1 414 395 361 063030 688 310 90 1 4 1 54 2815 521 53609 080 900 90 1 5 1 6252 1719 202 309047 379 960 90 1 625 2 03 4463 227 8312050 4311 060 90 2 0 2 830 5194 631 5515052 2712 050 89 Total100 033 881 4021055 4014 360 88 Tails100 0046 820 68 V Gupta M K Mohanty Int J Miner Process 79 2006 9 1712 desired product ash is higher than 6 73 incremental ash levels These iterative steps were continued until the calculated plant product ash converged to the desired product ash level The maximum plant yield was cal culated as the weighted average of individual circuit yieldsobtainedattheincrementalashcontent b 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Product Ash Mass Yield Mass Yield c Product Sulfur Heavy Medium Vessel Heavy Medium Cyclone Spiral Froth Flotation a 0 20 40 60 80 100 010203040506070 0510152025304035 0 80 911 11 21 31 41 61 5 8090 Incremental Ash Yield Heavy Medium Vessel Heavy Medium Cyclone Spiral Froth Flotation Heavy Medium Vessel Heavy Medium Cyclone Spiral Froth Flotation Fig 3 A graphical illustration of determining maximum yield by equalizing incremental ash from each cleaning circuit of a coal preparation plant V Gupta M K Mohanty Int J Miner Process 79 2006 9 1713 corresponding to the desired overall plant product ash content The incremental product quality approach has been proved to be an excellent plant optimization approach to maximize clean coal yield while dealing with only one quality constraint i e product ash However this approach becomes increasingly complex and also may lead to erroneous conclusions if more than one product quality constraints have to be satisfied simultaneously Incremental product ash which is also referred to as the instantaneous ash content that cannot be directly mea sured since it is the ash content of a material with a single precise density Abott and Miles 1990 Practi cally elementary ash content of a very close density fraction of a material is the best approximation of instantaneous ash Elementary quality of a product is referred to the grade of its dirtiest particle or particle s with respect to the specific quality being considered Based on the fundamental principle behind the incre mental product quality approach it may be intuitive to expect that the maximum yield grade curve for exam ple yield ash curve generated for an overall plant will be at its best when generated by equalizing the specific incremental quality incremental ash It is also quite understandable that the dirtiest particle s with respect to one assay for example ash content is not necessar ily the same particle s with respect to another assay for example sulfur content Therefore the maximum overall plant yield determined by equalizing the incre mental ash may be different from the maximum plant yield determined by equalizing incremental sulfur con tent obtained from each circuit Even in case the max imum plant yields obtained from both approaches are equal the combinations of individual circuit yields causing this may be completely different Thus main taining the required quality of the overall plant product with respect to both ash and sulfur will be difficult if one tries to achieve the plant yield maximized on the basis of both incremental ash and incremental sulfur contents This phenomenon is further illustrated for clarification u