1 LINEAR OPTIMIZATION-exercises using land use systems generated with LUCTOR H. Hengsdijk Introduction This document with its exercises is a brief introduction how technical coefficients generated with LUCTOR can be used in a linear programming model. This type of models is frequently used in land use studies (Rabbinge & Latesteijn, 1992; WRR, 1992; Veeneklaas et al., 1994). It goes beyond the scope of this document to fully explain the technical details and background of linear programming. Merely suffice to say here that in the linear programming exercises presented combinations of land use systems are selected that maximize the total profit (net return) under various boundary conditions (restrictions). For educational purposes simple optimization exercises have been developed in EXCEL8.0 with ready-made technical coefficients for a number of crops and management options (cropping systems). A set of scenarios with specific boundary conditions have been pre-programmed that can easily be executed by the user. The case study presented is for a (hypothetical) farm in the Neguev settlement in the Atlantic Zone (location 83 33’E and 10 12”N). The altitude is between 10 and 50 m above sea level in a region where climate can be classified as very humid, without dry months. The settlement originated when squatters occupied land of the hacienda Neguev in 1979. The IDA (Instituto de Desarollo Agropecuario) divided the Neguev into farms of 10, 15 and 17 ha (Schipper, 1996). Three soils have been classified in the Neguev: fertile well drained (SFW), infertile well drained (SIW) and fertile poorly drained soils (SFP). The latter soils are unsuitable to grow crops due to waterlogging. The total area of the case study farm is 10 ha and encompasses 5 ha of SIW and 5 ha of SFW soils. Although pastures and forests are still the major land uses in the region the case study will exclusively focus on crops. As part of a structural adjustment program, the Costarican government lowered prices of basic food crops (including maize and beans) which have pushed cropping towards palmheart and pineapple. Major concern in the region are low farm incomes, while at the same time land is used in an unsustainable way. I.e. farmers deplete their soils since insufficient fertilizers are used to maintain soil fertility. Moreover, biocides 27/06/2016 612930688 2 are frequently used, especially in pineapple cropping systems. Low credit availability limits farmers to invest in more productive and sustainable production methods. This case study will investigate the possibilities to improve the situation of farmers in the Neguev while taking simultaneous account of income, sustainability and environmental effects. LUCTOR was used to generate a number of land use systems. These land use systems are offered to a (simple) Linear-programming model that optimizes farmers profit (net revenue) given various restrictions related to sustainability, environmental and socioeconomic effects. The combination of the optimized objective (profit) and restrictions are called ‘goals’. Using the LP-model with different goals, so-called ‘scenarios’ can be created. The results of these scenarios quantify trade-offs that exist among various goals. Examples of scenarios include a farmers profit scenario, a sustainability scenario or an environmental scenario. In brief, the LP model presented finds the ‘optimum’ number and types of land use systems by maximizing ‘profits’, i.e. the sum of economic yield (i.e. sold products) minus the sum of all input and labor costs (for field operations). Physical restrictions are set on soil use (i.e. no more of each land unit can be used than the maximum available), on labor (no more labor can be used than available at the farm), on the area that each crop may comprise (to avoid soil born diseases, common in mono cropping systems), and - optionally - on sustainability, environmental or socio-economic parameters. Prices of inputs and outputs are those prevailing in the Atlantic Zone in mid 1996. Installation The exercises can be run on any type of computer with WINOWS95 and EXCEL. Copy the file LP_MOD.XLS to any directory on your hard disk C:\. Open EXCEL by pressing the EXCEL icon in your Microsoft toolbar or Initial start up. Press [Ctrl+o] to open the file LP_MOD.XLS. After two introduction dialog boxes the user may view the technical coefficients. The file consists of two worksheets with technical coefficients. The optimization model is hidden. Worksheet MODEL1 contains technical coefficients of actual land use systems only, while worksheet MODEL2 contains technical coefficients of both actual and alternative land use systems. Before starting the optimization exercises the user must press [Ctrl+x] to start a macro that links the EXCEL solver file to the exercises. The location of this file may be different on pc’s due to differences in installation of Microsoft office and/or WINDOWS95. When the SOLVER.XLA file is not available on your hard disk C:\ a dialog box appears to inform you 27/06/2016 612930688 3 that you first should install SOLVER.XLA using your Microsoft Office software. Without this file the exercises will not work. After you have installed SOLVER.XLA you have to repeat the opening procedure of LP_MOD.XLS and press [Ctrl+x] once again to link the file with the LP-exercises. The exercises using technical coefficients of actual land use systems may be activated pressing [Ctrl+y], while pressing [Ctrl+r] will activate the exercises using both actual and alternative land use systems. It is advised to start with the exercises using actual land use systems only. In this order the exercises will be discussed in the following. A last remark: The regional settings of your pc should separate decimals with a dot. Some of the exercises will show errors using comma separation of decimals. The English (United States) setting uses dot separation. Scenario 1: base scenario First, a so-called base-scenario is run using technical coefficients of actual land use systems only. This base scenario has no restrictions with regard to sustainability or environmental effects, i.e. profit is maximized under restrictions of available soils and labor availability. Soils available for each crop (crop rotation restrictions) have been set at 2 ha for each crop type, i.e. a maximum of 2 ha of the farm area may be cropped with cassava, 2 ha with beans, 2 ha with maize-grain and maize-fresh cobs, 2 ha with pineapple and 2 ha with palmheart. In this scenario 64 actual cropping systems are studied as shown in the worksheet MODEL1. For educational purposes only a selection of technical coefficients generated by LUCTOR are shown and one coefficient has been added, gross income (column heading GROSS) which is the value of the sold produce. The costs of production (column heading CCOST) exclude costs for fertilizers and labor. However, in the optimization of profit (or net revenue) fertilizer and labor costs are discounted for. Technical coefficients for actual cropping systems have been generated for maize-grain, maize-fresh cobs, cassava, pineapple for the local market and palmheart. Press [Ctrl+y] to open a dialog box with scenario options and select ‘base run’. Press [OK] when the scenario is selected. The LP model will start to optimize the net return objective in an iterative way. The optimization process stops when a solution is found that does not allow further progression of the objective function. While running the LP-model the screen will flicker now and then. When an optimal solution is found the LP-model returns with a bell sound and a message box with solver results. It says that the solver found a solution and that all constraints and 27/06/2016 612930688 4 optimally conditions are satisfied. The option button with ‘Keep solver solution’ is turned on. Press [Return] or click the OK button. The model continues with writing the objective values, selected land use systems and constraint values to the output worksheet base_run. When finished the base_run worksheet is the active sheet with results of the LP-model. View the contents of this worksheet. The first item at the top of the base_run worksheet is the so-called ‘objective function’, which is the variable that has been maximized. In our case study, this is the total profit of cropping systems: Total net revenue in col/year Net revenue SOLUTION 1619192 This net revenue is the highest profit that can be obtained with cropping systems at the farm since - except for the restriction on maximum soil use, the crop rotation restriction and the available labor that must be satisfied! - No restrictions on sustainability/environmental effects or credit availability were imposed. The base run can thus be interpreted as pure farmers’ profit scenario (short-term profit). Then follows the list of selected land use systems that resulted in the highest profit realized (see Annex I for the identification codes of land use system codes): Land use systems in ha SOLUTION SFW.ME.F0HHL 2.00 SFW.ZC.F0HHL 1.00 SFW.AM.F0HLH 2.00 SIW.BG.F0HLL 2.00 Apparently, it is most profitable to concentrate on cassava, pineapple and palmheart systems since these attain their maximum area. The fertile well-drained soils (SFW) are preferred to the infertile well-drained soils (SIW): Of the total cropped area 5 ha are SFWsoils while only 2 ha of SIW-soils (with pineapple) are selected. The preference for SFW soils can also be seen in the following section of the output, the part with restrictions. All the SFW soils are used while 3 ha of SIW soils are left fallow. 27/06/2016 612930688 5 Soil use in ha MAXIMUM SOLUTION SFW 5.00 5.00 SIW 5.00 2.00 This is not caused by a shortage of labor, which is shown by the labor restriction: Labor use in days/yr labor use MAXIMUM SOLUTION 300.00 228.90 The solution shows that less labor is required than the maximum available, which was set at 300 days per year. The crop rotation restriction shows that three crops attain their maximum allowable area (cassava, pineapple and palmheart). Since all SFW soils are used in the solution the LP-model can not select any other land use systems because maize and beans, which do not reach their limit, are unsuitable to grow at infertile well-drained soils (SIW). Crop rotation in ha MAXIMUM SOLUTION Cassava 2.0 2.0 Maize 2.0 1.0 Beans 2.0 0.0 Pineapple 2.0 2.0 Palmheart 2.0 2.0 On the other (costs and sustainability and environmental) indicators were no restrictions set in this scenario. Note that the 7 ha of cropped area result in a N-deficit of about 157/7 = 22 kg N/ha. The total amount of biocide (herbicides, fungicides and insecticides) used is about 12/7 = 1,71 kg ai/ha. Total farm costs in col/yr MAXIMUM SOLUTION Infinite 744857 N-balance in kg N Infinite -157 P-balance in kg P Infinite 27 K-balance in kg K Infinite -330 biocide use in kg a.i Infinite 12 Costs without fertilizer and labor Sustainability indicators 27/06/2016 612930688 6 The base run scenarios shows the optimal combinations of cropping systems given the boundary conditions imposed for an individual farm. When all farmers in a region would use the same cropping pattern over-production of certain products could lead to price drops while for other products (e.g. beans in this example) a shortage will result in a price increase. It is obvious that this kind of price effects will determine the optimal combination of selected cropping systems. Note also, that the selected production systems are highly unsustainable as expressed by the large amounts of soil nutrient depletion, which means that yields cannot be maintained on the long run. In the following exercise we will analyze the effect of fertilizers to counter balance the loss of nutrients. Fill-in some summary information in Table 1 (at the end of this document) for comparison with other scenarios (pre-filled-in as example). Scenario 2: Reduced N-depletion scenario In this scenario, the same model with its technical coefficients as in the base scenario is run, the only difference consists of a restriction on the N-balance. The base scenario showed that current-cropping practices without a restriction on the allowable soil nutrient depletion would result in a considerable exhaustion of soil nutrients. Ultimately this will result in declining yields and farm profits. The reduced N-depletion scenario expresses a hypothetical (political) wish to for a more sustainable agriculture by reducing soil depletion. In this scenario it is assumed that nutrient depletion is reduced by 50% compared to the base run. I.e. the maximum allowable N-depletion is set at –79 kg N for the entire farm. Press [Ctrl+y] to open a dialog box with scenario options and select ‘Reduced N depletion scenario ’. Press [OK] when the scenario is selected. The LP model will start to optimize the net return objective. The optimization process stops when a solution is found that does not allow further progression of the objective function. While running the LP-model the screen will flicker now and then. When an optimal solution is found the LP-model returns with a bell sound and a message box with solver results. It says that the solver found a solution and that all constraints and optimally conditions are satisfied. The option button with ‘Keep solver solution’ is turned on. Press [Return] or click the OK button. The model continues with writing the objective values, selected land use systems and constraint values to the output worksheet N_restriction_run. When finished the N_restriction_run worksheet is the active sheet with results of the scenario. 27/06/2016 612930688 7 View the contents of this worksheet and fill in Table 1 (at the end of the document). Note the differences with the base run (scenario 1), as well as the trade-offs between indicator values. Due to the N-restriction imposed the area of the most nutrient depleting crop (cassava) is decreased compared to the base run. Maize that does not contribute much to the farm profit is no longer selected. Compared to the base run a part of the cassava is produced at infertile well drained (SIW-land units) with lower yields and less negative N-balances compared to cassava systems at SFW land units. Palmheart systems have a positive N-balance due to the currently high N-gifts. Since N-losses are smaller at SFW land units than SIW land units, palmheart is shifted to SFW land units on which N-balances are more positive than on SIW land units. The net revenue decreases with 14% compared to the base run, indicating the ‘price’ of a more sustainable system (however, still not 100% sustainable!). It is emphasized that in this scenario only actual production systems are analyzed. The effect of introduction of alternative land use systems, which are aimed at equilibrium nutrient balances, will be analyzed in scenario 4, 5 and 6. Scenario 3: Reduced biocide scenario The same model as in the base scenario is run, with a restriction on the use of biocides as quantified by BIOA in the technical coefficients. The scenario expresses a hypothetical (political) wish to decrease the use of biocides by 50% from the base situation. Therefore, a restriction on BIOA is set as 50% of 12 kg ai = 6 kg a.i for the total farm area. Press [Ctrl+y] to open a dialog box with scenario options and select ‘Reduced biocide scenario ’. Press [OK] when the scenario is selected. When the optimization procedure is finished save the results by clicking OK button or pressing [Return]. The model results are written to the worksheet biocide_run. When finished this worksheet is the active sheet with results of the scenario. View the contents of this worksheet and fill in Table 1 (at the end of the document). Note the differences with the base run (scenario1), as well as the trade-offs between indicator values. Reduction of the use of biocides is realized by selecting low herbicide systems. Note that the low pesticide option is not chosen. Apparently, the yield reduction due to low pesticide use is too large to be profitable. To satisfy the biocide restriction the pineapple area is 27/06/2016 612930688 8 reduced. Pineapple is the most demanding crop concerning biocides. This allows shifting a part of the palmheart towards more productive land units (SFW). The total area is reduced to 5.4 ha while the net revenue is reduced with 16% compared to the base run. Note that the costs of production are much lower. Why? Scenario 4: Introduction of alternative land use systems scenario In this scenario a completely new set of alternative land use systems is added to the former set with only actual cropping systems, a total of 128 land use systems, see worksheet MODEL2. To study the feasibility of using fertilizer to compensate for soil nutrient depletion (and thus to produce in a sustainable way) the same crops with the same management options were generated with LUCTOR as for actual land use systems used in the previous scenarios. The production level of alternative land use systems was set at 100% of the maximum attainable level and are therefore higher than the current attained yields. View briefly the model-worksheet and recognize that alternative land use systems have been added to the technical coefficients compared to the previous scenarios (compare worksheet MODEL1 and MODEL2). Press [Ctrl+r] to open a dialog box with scenario options and select ‘Introduction of alternative systems scenario ’. Press [OK] when the scenario is selected. The LP-model optimizes the net return of the farm in the same way as in the previous scenarios. When an optimal solution is found press the OK button or [Return]. The model writes the results to the worksheet alternative_run, which will be the active worksheet when finished. View the contents of this worksheet and fill in Table 1 (at the end of the document). Note the differences with the previous scenarios, as well as the trade-offs between indicator values. Although the costs (without labor and fertilizer costs) of alternative land use systems are twice as high as those of actual land use systems, it is profitable to choose for alternative land use systems. The net revenue can be tripled by selecting high yielding alternative land use systems compared to the base run of scenario 1. In terms of crop choice this scenario resembles the base run scenario: cassava, pineapple and palmheart are the most preferred crops that each attain their crop rotation constraint of 2 ha. Note that labor availability in this scenario limits further increase of the net revenue. Note also the high input of biocides required to realize the high yields of alternative land use systems. 27/06/2016 612930688 9 Scenario 5: Reduced credit scenario The previous scenario showed that farm profit can be increased considerably through the introduction of alternative (and 100% sustainable!!) land use systems. However, costs for machinery, biocides, plant material etc. are doubled compared to the base run (scenario 1). These investments are even much higher as the costs for fertilizer and labor, - which are substantial for alternative land use systems and which are not accounted for in the costs as shown!-, are taken into account. It is unlikely that a farmer has the financial sources to realize these investments. In this scenario it is assumed that the farm has a credit restriction at the amount of the costs of the first scenario, i.e. 744857 col. for the entire farm (see results scenario 1 in Table 1). This is almost half of the investments required in scenario 4. Press [Ctrl+r] to open a dialog box with scenario options and select ‘Reduced credit scenario ’. Press [OK] when the scenario is selected. The scenario results are written to the worksheet credit_run. When finished this worksheet is the active sheet with results of the scenario. View the contents of this worksheet and fill in Table 1 (at the end of the document). Note the differences with scenario 4, as well as the trade-offs among indicator values. Note that net return drops almost 25% while investment costs are 50% lower than in scenario 4. The production has shifted towards less expensive (less input demanding) cropping systems. The area with pineapple (the most expensive crop) is sharply reduced while for all land use systems the low herbicide option is preferred since these are a little bit more profitable than the high herbicide options. These low herbicide options require fewer herbicides but more labor, which is also shown by the reduction in biocide use with 45%, compared to scenario 4. Apparently, the credit constraint is restricting the objective value more than the labor constraint. In the previous scenario (scenario 4) only labor was restricting a further profit increase. Therefore, in that scenario labor extensive management options were selected with high herbicide and some high mechanization options since these require less labor (see results scenario 4 in Table 1). More area could be taken in production, 6.7 ha compared with 5.92 ha in the credit scenario, which allowed raising farm profits. Scenario 6: Reduced labor availability scenario In this scenario, the same model as scenario 4 is run, the only difference consisting of a restriction on the availability of labor. It is assumed that the farmer can work a part of his time 27/06/2016 612930688 10 at a nearby banana plantation reducing his total labor availability at the farm at 200 mnd per year (compared to the 300 mnd per year in other scenarios). Press [Ctrl+r] to open a dialog box with scenario options and select ‘Reduced labor availability scenario ’. Press [OK] when the scenario is selected. The scenario results are written to the worksheet labor_run. When finished this worksheet is the active sheet with results of the optimization procedure. View the contents of this worksheet and fill in Table 1 (at the end of the document). Note the differences with scenario 4, as well as the trade-offs between indicator values. Note that the net revenue is reduced with 17% compared to scenario 4, although the labor availability is reduced with a third. Because the labor availability is limited in this scenario, only two land use systems are selected both with high input of capital goods (biocides and mechanization) which require relatively little labor. With the little labor available the farmer cannot cultivate much has, even a part of the fertile well-drained soils are taken out of production. When finished with scenario 6, close the file LP_MOD.XLS. Do not save the file. Conclusions The results of the scenarios (as summarized in Table 1) indicate the ‘playground’ for sustainable development of this particular farm type. By running different scenarios, tradeoffs between socio-economic, sustainability and environmental objectives are quantified. The remarks made at the end of scenario 1 are valid for all other scenarios; I.e. at an aggregate level of analysis, e.g. a region, product prices will become endogenous and will affect the outcome of a scenario. The highest profits are realized with sustainable (alternative) land use systems as is shown in the last three scenarios. The higher yields that can be attained with a well-balanced mix and high level of inputs (and associated high costs!) will pay off. However, The selected land use systems do not account for yield risks caused by e.g. adverse weather conditions. Moreover, it assumes that farmers timely apply the required inputs and carry out the required operations. This requires a well-educated and precise farm management. Based on the current production systems a reduction in the use of biocides or in N-depletion of 50%, both have a ‘financial trade-off’ of about 15% on farmers profit (while still using unsustainable cropping systems), as is shown in scenario 2 and 3. 27/06/2016 612930688 11 Table 1 Summary of scenario results. SCENARIO: Base scenario Net revenue 1619192 Land use systems (type +ha) SFW (ha) SIW (ha) labor (mnd/yr) costs (col/yr) N-balance (kg N/ha) Biocide (kg ai/ha) SFW.ME.F0HHL = 2 SFW.ZC.F0HHL = 1 SIW.AM.F0HLH = 2 SIW.BG.F0HLL = 2 5 2 228 744857 Reduced N depletion scenario Reduced biocide scenario Introduction of alternative systems scenario Reduced credit scenario Reduced labor availability scenario 27/06/2016 612930688 -157 12 12 27/06/2016 612930688 13 Annex I Technical Coefficients of land use systems Description of technical coefficient Unit Code Export quality product kg/ha/yr EXP Domestic quality product kg/ha/yr DOM Refuse product kg/ha/yr REF Labor requirements – discounted annuity mnd/yr CLABA Labor requirements – mean mnd/yr CLABM Costs col/yr CCOST N-balance kg N/ha/yr NBAL P-balance kg P/ha/yr PBAL K-balance kg K/ha/yr KBAL Volatilisation kg N/ha/yr NVOL Denitrification kg N/ha/yr NDEN N-Leaching kg N/ha/yr NLEA Biocide index BIOI Biocide use kg ai/ha/yr BIOA Fertilizer N-input kg N/ha/yr FIN Fertilizer P-input kg P/ha/yr FIP Fertilizer K-input kg K/ha/yr FIK P-fixation kg P/ha/yr PFIX K-leaching kg K/ha/yr KLEA Tractor use trdays/yr TRAC Sowing equipment use seqdays/yr SOWE Multicultivator equipment use meqdays/yr MULT Fertilizer spreader use feqdays/yr FEQ Knapsack sprayer use keqdays/yr BOMB Seedlings/seed #/ha or kg/ha SEED 27/06/2016 612930688 14 Annex II Identification codes of land use systems Land use systems have an identification code, for example: SFP.ZM.F9HHL code 1: Land units can be: SFW = Soil fertile well drained SIW = Soil infertile well drained SFP = Soil fertile poorly drained code 2: Crop type can be: AM = pineapple for local market BG = palmheart ME = cassava PV = beans ZM = maize grain ZC = maize fresh cobs code 3: Combination of type of land use system, target yield level, pesticide level, herbicide level and mechanization level. First letter plus number: type of land use system plus yield level. F0 stands for actual land use systems, F1…F9 stand for alternative land use systems in which F9 is the maximum level and F1 is 10% of this level. Second letter: pesticide level (L = low, H = high) Third letter: herbicide level (L = low, H = high) Fourth letter: mechanization level (L = low, H = high) 27/06/2016 612930688