Datasets para Problemas de Lote e Programação na Produção de Bebidas à Base de Frutas

Data in Brief 35 2021 106810 Contents lists available at ScienceDirect Data in Brief journal homepage wwwelseviercomlocatedib Data Article Datasets for lot sizing and scheduling problems in the fruitbased beverage production process Juan Piñeros a Alyne Toscano b Deisemara Ferreira c Reinaldo Morabito d a Department of Production Engineering Federal University of São Carlos SorocabaSP Brazil b Department of Production Engineering Federal University of Triângulo Mineiro UberabaMG Brazil c Department of Physics Chemistry and Mathematics Federal University of São Carlos SorocabaSP Brazil d Department of Production Engineering Federal University of São Carlos São CarlosSP Brazil a r t i c l e i n f o Article history Received 8 December 2020 Revised 22 January 2021 Accepted 26 January 2021 Available online 29 January 2021 Keywords Fruitbased beverage Lot sizing and scheduling problems Temporal cleaning Pasteurized juice Sequence dependent setups Production planning a b s t r a c t The datasets presented here were partially used in Formu lation and MIPheuristics for the lot sizing and scheduling problem with temporal cleanings Toscano A Ferreira D Morabito R Computers Chemical Engineering 1 in A decomposition heuristic to solve the twostage lot sizing and scheduling problem with temporal cleaning Toscano A Ferreira D Morabito R Flexible Services and Manufactur ing Journal 2 and in A heuristic approach to optimize the production scheduling of fruitbased beverages Toscano et al Gestão Produção 2020 3 In fruitbased produc tion processes there are two production stages preparation tanks and production lines This production process has some processspecific characteristics such as temporal cleanings and synchrony between the two production stages which make optimized production planning and scheduling even more difficult Thus some papers in the literature have pro posed different methods to solve this problem To the best of our knowledge there are no standard datasets used by re searchers in the literature to verify the accuracy and perfor mance of proposed methods or to be a benchmark for other researchers considering this problem The authors have been DOI of original article 101016jcompchemeng2020107038 Corresponding author Email address alynetoscanouftmedubr A Toscano httpsdoiorg101016jdib2021106810 23523409 2021 The Authors Published by Elsevier Inc This is an open access article under the CC BY license httpcreativecommonsorglicensesby40 2 J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 using small data sets that do not satisfactorily represent dif ferent scenarios of production Since the demand in the bev erage sector is seasonal a wide range of scenarios enables us to evaluate the effectiveness of the proposed methods in the scientific literature in solving real scenarios of the problem The datasets presented here include data based on real data collected from five beverage companies We presented four datasets that are specifically constructed assuming a scenario of restricted capacity and balanced costs 2021 The Authors Published by Elsevier Inc This is an open access article under the CC BY license httpcreativecommonsorglicensesby40 Specifications Table Subject Management Science and Operations Research Specific subject area Production Planning Lot Sizing and Scheduling TwoStage Production Temporal Cleaning Fruitbased Beverage Type of data Tables Images Charts Graphs Figures How data were acquired The data were acquired through field research observation documents provided by companies as well as computer generators using different probability distributions Data format Raw Analyzed Filtered Parameters for data collection The data were collected through guided tours to fruitbased beverage companies During these visits unstructured interviews were conducted with decision makers and employees Description of data collection Initial data were obtained from field research conducted in five companies from the fruitbased beverage industry in Brazil One of them is a multinational company The data were collected observing their production processes using electronic spreadsheets provided by some of these companies and through interviews conducted with production managers and employees from production lines as well as production planning and control departments From these initial data other instances were generated by computer generators using different probability distributions These distributions and their respective parameters were defined according to the initially collected data Some parameters were still adjusted aiming to have various realistic scenarios Data source location CityTownRegion São Paulo State Minas Gerais State Country Brazil Data accessibility Repository name Mendeley data Data identification number 1017632j2 3 gbskfw1 Direct URL to data httpsdatamendeleycomdatasetsj2x3gbskfw1 Related research article Toscano A Ferreira D Morabito R Formulation and MIPheuristics for the lot sizing and scheduling problem with temporal cleanings Computers Chemical Engineering Available online 2020 Doi 101016jcompchemeng2020107038 Toscano A Ferreira D Morabito R A decomposition heuristic to solve the twostage lot sizing and scheduling problem with temporal cleaning Flexible Services and Manufacturing Journal 31 2019 142173 Doi 101007s10696 017 9303 9 Toscano A Ferreira D Morabito R Trassi M V C A heuristic approach to optimize the production scheduling of fruitbased beverages Gestão Produção 274 e4869 2020 httpsdoiorg1015900104 530x4869 20 Value of the Data These datasets are useful because they are based on real data Thus they illustrate a more practical perspective of the problem and its complexities Moreover due to their variation in costs and capacity parameters they represent different scenarios of real companies It is well known that there is a lack of this kind of data in the scientific literature On the other hand research based on more realistic data has been increasingly demanded J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 3 This data set consists of a collection of instances for the classic problem of lot sizing and scheduling problems with sequencedependent setup timescosts and changeover times sat isfying the triangular inequality In addition to these characteristics there are some additional parameters such as temporal cleaning which characterize the production processes with the pasteurization step Therefore studies related to the fruitbased beverage industry or other food industries with similar production process features for example milk production pro cesses can benefit from this collected data Besides the parameters can be used to generate instances based on real data for other lot sizing and scheduling problem classes Part of the datasets was used in research concerning a fruitbased beverage industry There fore the datasets provide instances for future research comparing the results and establishing a new benchmark for the problem 1 Data Description Obtaining real instances to validate methods proposed in the scientific literature still poses a challenge 4 However according to 5 there is a trend and a research opportunity in lot sizing and scheduling applied to real problems Therefore describing real data in detail is an important contribution to future research The dataset presented in this article describes the main parameters of a fruitbased beverage production process which are important and should be considered when carrying out produc tion planning and scheduling The fruitbased production process consists of two main stages In Stage I raw material is mixed with water in preparatory tanks The beverage produced in Stage I is pasteurized and filled in Stage II called Line thus generating the final items These stages are dependent and must be synchronized in production planning and scheduling Each tank in Stage I is dedicated to a production line in Stage II In order to change over items cleaning is required The times and costs for this cleaning can be sequencedependent or independent In these production processes there are manda tory cleanings also called temporal cleanings which are necessary when the time spent from completing the last cleaning reaches a permitted maximum time without cleaning In the bev erage production process the changeover times respect the validity of the triangular inequality and for the fruitbased production process this characteristic is also true Therefore production planning and scheduling consist of deciding the quantity of each item that must be produced in each tankline in each period and the production sequence of these items to meet the demand minimizing backorder and inventory items as well as the times and costs spent on temporary cleaning and cleaning by item changeovers For more details about these production processes see 13 To create the dataset of instances field research was conducted in five companies from the fruitbased beverage industry in Brazil one of which is a worldwide brand We present a total of 92 instances divided into 4 groups G1 G2 G3 and G4 of 23 instances each These instances consist of real and generated data based on real data Although the 23 instances of the G1 group vary in size and the data are based on real in formation some parameter values such as costs and capacity do not vary to represent some important scenarios For example scenarios with more restricted capacity or scenarios where inventory and backorder costs are more balanced than changeover costs To compose a set of more representative instances we then generated 3 more groups of instances G2 G3 and G4 The G2 group instances consist of the G1 instances with a reduction in time capacity by 10 The G3 group instances were created reducing inventory and backorder costs of the G1 instances The G4 group instances were created with modifications made in groups G1 and G2 Thus an instance is always named by the group to which it belongs followed by the position that it occupies in the group For example instance G115 is the fifteenth instance of Group 1 All parameters that comprise an instance are described in Table 1 All instances are presented in electronic spreadsheets in the repository httpsdatamendeleycomdatasetsj2x3gbskfw1 More details about each group are described in the next section 4 J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 Table 1 Parameters of the instances Parameter Description J M T Total number of items tankslines and time periods respectively J Set of items from 1 to J M Set of tankslines from 1 to M Each line has a dedicated tank T Set of periods from 1 to T I j0 I j0 Initial inventory backorder level for item j J in units t c I t c II Time spent to perform a temporal cleaning in a tank in a line in minutes t e I max t e II max Maximum time elapsed from the last temporal cleaning in a tank in a line in minutes d jt Demand for item j J period t T in units ρj Quantity of beverage in one unit of item j J in liters ca p mt Available capacity time of tankline m M in period t T in minutes b I ij b II ij Changeover times from item i to item j in stage I in stage II in minutes u b j l b j Maximum minimum production quantities for the lot sizes of item j J in the tanks in liters pt Beverage production time of one lot in a tank considered independent from the lot size in minutes s m Time spent for line m M bottling one liter of beverage in minutes h j h j Nonnegative inventory backorder cost for one unit of item j J c ij Cost of process changeover from item i to item j in the production process ct Temporal cleaning cost O mt Ordered set of available lots batches of the same item for production in tankline m M in period t T Q I mt Q II mt Ordered set of available temporal cleanings for tank line m M in period t T O mt Maximum number of lots of the same item that can be produced in tankline m M in period t T Q I mt Q II mt Maximum number of temporal cleanings for tank line m M in period t T 2 Experimental Design Materials and Methods In the data collection process several visits were made to 5 fruitbased beverage companies from 2015 to 2019 The field research consisted of several guided tours During these visits a production manager from each factory described the entire process of producing fruitbased bev erages In other unstructured interviews data were provided via some electronic spreadsheets and via information provided by the production planner This material was then organized to generate the instances described in this paper The instances of group G1 are detailed in Section 21 The other instances of groups G2 G3 and G4 are derived from the G1 instances and are explained in Section 22 These G1 group instances were used by 1 21 G1 group of instances The G1 group consists of the first instances generated for the problem For each instance some parameters were real data and others were generated based on real data In Table 2 for each parameter the type of data real or generated for each instance is specified from 1 to 23 211 Real data In the companies visited an item is considered a sixpack with 6 bottles of a certain bev erage The total number of items J varied in the set 3 4 5 6 10 15 20 the total number of preparation tankslines M varied in the set 2 4 6 and the total number of periods T varied in the set 2 4 5 6 These data are real for all instances Table 2 The size of each instance was determined by the number of items tankslines and periods They are shown in Table 3 J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 5 Table 2 Classification of parameters in real data or based on real data for G1 instances Param Instance Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 J M T I j0 I j0 t c I t c II te I max te II max d jt ρj ca p mt b I ij b II ij l b j u b j pt s m h j h j c ij ct O mt Q I mt Q II mt Real data Data generated based on real data 6 J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 Table 3 Number of items tankslines and periods for G1 instances Instance Number Parameter 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 J 3 3 3 3 3 4 5 5 5 5 5 5 6 10 10 10 10 15 15 15 20 20 20 M 2 2 2 2 2 2 2 2 2 2 2 2 2 2 4 4 4 6 6 6 6 6 6 T 2 2 2 2 2 4 4 4 4 4 4 5 4 4 4 4 4 4 4 4 6 6 6 Among the 5 companies visited 3 of them can be considered small and medium sizes mainly due to the number of items produced and machines available for production and the volume of production They are represented in Table 3 by instances 1 to 6 The other 2 companies are large and are represented by instances 7 to 23 We also observed in the field research that the initial inventory and backorder levels for all items were always zero that is I j0 0 and I j0 0 for all instances The time spent on perform ing a temporal cleaning was 50 min for stage I and 300 min for stage II in all companies visited Thus these values are real for all instances t c I 50 and t c II 300 The same occurs for the maximum time elapsed since the last temporal cleaning in the line and tank These values were identical for all companies and therefore they are real data for all instances with te I max 1 445 and te II max 2 885 min respectively It can be observed in Table 2 that most parameters of instances G17 G18 and G112 are real data These three instances were collected from a worldwide brand of a fruitbased beverage whose plant is located in São Paulo State Brazil For these instances we collected the demands of five items orange passion fruit grape strawberry and pineapple Instances G17 G18 and G112 correspond to one month each nonconsecutive divided into weeks periods Notice that the month of instance G112 has 5 weeks For all items from these instances the quantity of beverage in liters to produce one unit of any item j J is ρ j 2 4 For these instances the available capacity time for each tankline m M in each period t T is 8550 min For instances G17 G18 and G112 the changeover times from item i to item j in both stages b I ij and b II ij are fixed and sequenceindependent This fact was observed only in this company For the other companies these times are sequencedependent and are presented in the next section These instances are the same real instances used by 2 and 3 We also col lected the lower and upper bounds for the lot sizes l b j and u b j respectively defined by the physical capacities of the tanks The time spent to produce the beverage in Stage I pt and the bottling time of one liter in line m M s m were also collected For more details of these parameter values please see the Supplementary Material 212 Estimated data based on interviews Not all parameters were provided by the companies Some of them were generated based on real information collected in the companies visited For example some parameters were ran domly generated using probability distributions such as the normal and uniform continuous dis tributions These parameters and the respective probability distributions are shown in Table 4 The variation intervals for each parameter were defined according to the information and data obtained in the five companies visited For parameters d jt ρ j u b j l b j and s m the collected samples were small Due to the experi ence and suggestion of the production planner interviewed we used the uniform distribution to generate these parameters which proved to be a reasonable option for this purpose The limits of the generation interval for each one of these parameters were defined based on the collected data and on the decisionmakers knowledge In some cases from more significant samples for the changeover times b I ij b II ij and the production times of a batch p t we computed descriptive statistics the mean and standard de viation and visually verified that the data approached a symmetric distribution For this reason and based on the recommendation of the interviewed decisionmaker we consider it reasonable Table 4 Parameters generated using probability distributions Parameter Distribution ρj U09 12 dαt Uminρji maxρji dβt Umaxρji 2ρj maxρji djt U0 limjtρj t T j J α β blij N30 52 blij N150 302 ubj U10000 20000 lbj θubj and θ U020 06 pt N100 102 sm U40 200 Table 5 Penalties of the production process Parameter Instances G17 G18 G112 Instances G11 to G16 G19 to G111 G113 to G123 hj 10 U10 20 hj 100 10 hj cij 1 10 blij blijmaxblij blijijJi0 ct 1 1 J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 9 Table 6 Production lots for the production plan of instance G17 Flavor t 1 t 2 t 3 t 4 Number of lots Liters per lot Number of lots Liters per lot Number of lots Liters per lot Number of lots Liters per lot m 1 m 2 m 1 m 2 m 1 m 2 m 1 m 2 Orange 2 50 0 00 0 1 448333 1 399999 1 386666 19 266666 1 257333 1 125000 Passion 1 50 0 00 0 3 50 0 00 0 Fruit 1 266666 1 433333 2 250 00 0 4 250 00 0 Grape 2 416708 1 416708 8 416708 5 416708 12 266666 11 216666 2 40 0 00 0 1 325000 8 208333 1 409166 14 416708 1 416708 2 386666 1 309125 4 266666 20 216666 4 208333 1 208333 1 321416 20 216666 1 232625 8 208333 1 386666 1 314166 34 266666 4 208333 1 386666 1 213125 11 208333 1 286416 Strawberry 3 50 0 00 0 5 50 0 00 0 4 50 0 00 0 4 50 0 00 0 1 40 0 00 0 19 266666 13 216666 1 453333 1 428333 1 486666 1 423166 1 340 00 0 2 50 0 00 0 14 50 0 00 0 1 335166 1 397500 1 266666 1 166666 1 386666 1 40 0 00 0 1 264166 6 266666 1 325000 5 166666 23 266666 1 406666 1 247500 7 166666 10 216666 1 257333 1 179000 1 210666 1 166666 1 166666 1 166666 1 386666 Pineapple 2 50 0 00 0 6 50 0 00 0 1 453333 1 315333 2 266666 4 250 00 0 7 250 00 0 Table 7 Parameters that are identical and different in G1 G2 G3 and G4 groups of instances Parameter Group G1 G2 G3 G4 J M T Ij0 Ij0 tcI tcII teImax teIImax djt ρj capmt blij blij lbj ubj pt sm hj hj cij ct Omt QImt QIImt N Original Changed J Piñeros A Toscano and D Ferreira et alData in Brief 35 2021 106810 11 Period t 1 Tank 1 Line 1 Tank 2 Line 2 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 Period t 2 Tank 1 Line 1 Tank 2 Line 2 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 Period t 3 Tank 1 Line 1 Tank 2 Line 2 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 Period t 4 Tank 1 Line 1 Tank 2 Line 2 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 First period cleaning Temporal Cleaning CIP Flavor Changeover Waiting Orange Passion fruit Grape Strawberry Pineapple Fig 1 Gantt chart for the production plan of instance G17 QmtIIcapmtteImax 7 Therefore set QmtI ranges from 1 to QmtI and set QmtII ranges from 1 to QmtII All parameters described above for each instance are available in spreadsheets in the Supplementary Material 22 Generation of G2 G3 and G4 groups of instances In order to evaluate the instances generated for the G1 group we created a manual production plan for instance G17 In this G17 solution the demands of all items are met in all periods that is there are no backorders and inventories in this plan Table 6 presents the number of lots and the corresponding liters for each item that must be produced in each machine and period Notice for example that in machine m 1 of period t 1 52 lots of the Grape item and 18 lots of the Strawberry item are planned to be produced The Gantt chart for this production plan is shown in Fig 1 In Fig 1 the horizontal line is the production timeline and each rectangle indicates the time spent on producing a lot or waiting or cleaning etc It can be observed that there is available capacity in almost all machines and periods and the demand is met without backorders 12 J Piñeros A Toscano and D Ferreira et al Data in Brief 35 2021 106810 and inventories For example in line m 2 of period t 3 the production finishes in less than 20 0 0 min still leaving more than 6550 available minutes of the production capacity This fact suggests that we can decrease the available capacity ca p mt and obtain instances with more re stricted capacity Therefore we generated instances from group G2 reducing the value of ca p mt for all m M and t ϵ T of each G1 instances by 10 ie G2 instances are based on G1 in stances except for ca p mt In G1 group instances the inventory and backorder costs are much more penalized than the changeover cost as these are unit costs However for different companies it is more impor tant to meet the demand required for the period either with inventory or production during this period than not meeting the customer demand in the required period and losing cred ibility with that customer To represent these situations the G3 group instances were cre ated reducing inventory and backorder costs of the G1 instances with h j U 0 5 2 and h j 20 h j Max c ij for all j J respectively To obtain even more realistic scenarios the G4 instance group was also created with modifications made in groups G1 and G2 ie they have the value of ca p mt reduced by 10 and h j U 0 5 2 and h j 20 h j Max c ij Table 7 presents the parameters that are identical to G1 and the parameters that have been changed for each group A summary of the characteristics of the four groups of instances G1 G2 G3 and G4 is pre sented in Table 8 Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal rela tionships that could have appeared to influence the work reported in this paper Acknowledgments The authors would like to thank the Brazilian fruitbased beverage companies for their col laboration with this research This work was supported by the São Paulo Research Foundation FAPESP under Grant number 2016018601 the Brazilian National Council for Scientific and Technological Development CNPq under Grants numbers 43926420189 0404720180 and 30961220187 and the Coordination for the Improvement of Higher Education Personnel Brazil CAPES Finance Code 001 References 1 A Toscano D Ferreira R Morabito Formulation and MIPheuristics for the lot sizing and scheduling problem with temporal cleanings Comput Chem Eng 142 2020 107038 doi 101016jcompchemeng2020107038 2 A Toscano D Ferreira R Morabito A decomposition heuristic to solve the twostage lot sizing and scheduling prob lem with temporal cleaning Flexible Serv Manuf J 31 2019 142173 doi 101007s10696 017 9303 9 3 A Toscano D Ferreira R Morabito MVC Trassi A heuristic approach to optimize the production scheduling of fruitbased beverages Gestão Produção 27 4 2020 e4869 doi 1015900104 530X4869 20 4 A Clark B AlmadaLobo C Almeder Lot sizing and scheduling industrial extensions and research opportunities Int J Prod Res 499 2011 24572461 doi 101080002075432010532908 5 k Copil M Wörbelauer H Meyr H Tempelmeier Simultaneous lotsizing and scheduling problems a classification and review of models OR Spectrum 39 2017 164 doi 1010 07s0 0291 015 0429 4 6 TA Baldo MO Santos B AlmadaLobo R Morabito An optimization approach for the lot sizing and scheduling problem in the brewery industry Comput Ind Eng 72 2014 5871 doi 101016jcie201402008 7 FMB Toledo VA Armentano A lagrangeanbased heuristic for the capacitated lotsizing problem in parallel ma chines Eur J Oper Res 175 2006 10701083 doi 101016jejor200506029