«Embargo information No embargo required Item downloaded from Downloaded on 2016-05-23T13:09:50Z Thesis to obtain the ...»
structure able to cope with a variety of dairy cases, and to construct a base scenario as close as possible to each given case. The base scenario was simulated in order to compare its outputs to the figures known by the processor. Calibrations were made if the processor estimated that certain results were not representative and/or if a lack of consistency was detected.
Once a satisfactory representation of the manufactured process was achieved, the construction of alternatives scenarios jointly with the processor began. Building alternatives scenario included modifications in (i) processor’ current portfolio towards higher value products; (ii) the volume of milk collected per day and (iii) the payments to his milk suppliers. Outputs from these alternative scenarios were discussed and the support process was evaluated with the processor in a final meeting.
4.2. Synthesis of Results 4.2.1. About the simulation tool DairyPlant is based on the analysis of the milk supply from farm level to plant gate and the manufacturing process (Figure 11). DairyPlant was designed considering the volume of milk collected per dairy processor in a day. Each milk supplier (an individual farmer, a group of farmers or a private collector) is characterized by (i) milk quantity supplied to the dairy processor in a day; (ii) values of up to three quality components (milk composition or milk hygienic values) from each milk supplier; and (iii) farmers’ capacities to increase, decrease or keep their current quality levels if the payment system is changed. This capacity is subjectively assessed by the processor based on the knowledge he has of his suppliers, since there is no direct mathematical relation between the variation of the payment system and the modification of milk quality supplied by each farmer.
Processing analysis take values of total raw milk collected and quality components evaluated from the supply analysis to calculate the outputs of the dairy manufacturing process. The proportion of milk used in the production of each dairy product is selected by the software user according to the product portfolio selected for a given scenario. The list of product manufactured allows the introduction of
each dairy product, i.e. the quantity of milk or intermediate product required to produce 1 kg of dairy product, is also defined by the software user based on existent formulas or in-situ controlled experiments. Processing costs are split into milk collection, product-related processing, packaging and marketing costs. Each fixed cost is also defined and split between processed products according to each processor’s choice. At the end of the simulation, processors obtain the total profits related to a given dairy portfolio. Up to 10 marketed dairy products and 10 intermediate processed products can be included in scenarios.
Figure 11: DairyPlant processing model showing milk supply and transformation processes DairyPlant also allows the design of milk payment systems. It includes a milk base price plus a combination of up to three quality variables, either chemical or hygienic, assuming that these variables are actually measured at the plant gate and so defined for each supplier. For each variable the user gives the base value and a penalty and/or bonus for each point respectively above or below the base value. So, simulations may include payment systems with (i) only bonuses and no penalties; (ii) both;
gross product for the suppliers can then be carried out according to the quality supplied by each one to the plant. DairyPlant was developed using Microsoft Excel 2010, in a user-friendly way in order to facilitate its manipulation and understanding by the stakeholders involved in the support process. The
structure of DairyPlant consists in three modules (Table 13) as follows:
Table 13: Commented list of the variables included in DairyPlant
given plant: range of dairy products, milk quantity and quality per famer, payment system, costs and gross product per dairy product) and the results (calculated variables for the given case: plant profit and farmers’ gross products). Each scenario is run for one day considered as representative of the plant business throughout the year.
Parameters module includes variables required in the “input” module for characterizing a given scenario. The user also defines the raw material used and the quantity of raw material required to produce one kg. of a given processed product. These processing yields are essential to determine the amount of dairy products produced by a processor, based on both the collected milk quantity dedicated to each dairy product and its average quality of all the daily deliveries. The “input” module is divided into four sub-modules: Plant scenario (1 sheet), Supply (1 sheet), Fixed costs (1 sheet) and marketed products (10 identical sheets, one per product). In the “Plant Scenario” the user enters a base price, and optionally a base value (%) for up to three milk components and the economic value of each point respectively higher (bonus) or lower (penalty) than the base value (Figure 12). The “supply” sheet regroups information regarding suppliers’ daily milk delivery. For each of them, the user enters his daily volume and the milk quality values for the 3 components selected in the “Plant Scenario” sheet. The total volume of milk collected in a day and the weighted average values of milk quality composition are then calculated for the plant. Quality improvement is qualitatively defined by the processor based on his knowledge about his suppliers’ behavior regarding milk quality management.
In the “Fixed costs” sheet are entered all the costs which are independent from the quantity of milk collected per day. Finally, each of the 10 “marketed product” sheet represents the daily processing of one dairy product. Once the scenario is characterized, DairyPlant simulates the corresponding milk processing of the dairy plant. Results are presented in two separated sheets called “Plant results” and “Farmer results” respectively, based on (i); the total production costs (fixed and variable) linked to the dairy plant operation; the adjusted milk price corresponding to each farmer after bonuses and penalties and his gross product according to his quantity of delivered milk. This gross product corresponds to the milk cost for the processor.
4.2.2. Supporting small-scale dairy plants in selecting market orientations and milk payment systems As it was observed in the previous chapters, small-scale processors at Mantaro Valley face problems collecting uniform milk volumes through the year due to high competition for milk supply. Dairies’ capacity to deal with these constraints is limited, since they do not have sufficient information for selecting the most beneficial resource mix or for designing different milk payment systems. DairyPlant was tested with two small-scale dairy plants (DP1 and DP2) to show them potential benefits they could expect from modifying their current portfolio or from adopting a payment system based on milk quality.
Moreover, this approach attempted to develop a prospective thinking about milk quality on small-scale dairy processors, since they currently neglect the importance of rewarding their milk suppliers and managing milk quality on their manufacturing process. Scenarios were configured and the whole set, including the reference set, was simulated for each processor.
22.214.171.124. Product portfolio Fresh cheese is the main manufactured product of Peruvian dairies. It represents between 70 -80% of the milk processed, but it does not necessarily provide the highest profits. Varios scenarios based on
processors were interested in analyze the possibility to process dairy products with higher market value. The scenarios varied according to the amount of milk processors were able to reduce from fresh cheese towards other dairy products and the total profit expected at the end. Results showed that both small-scale dairy processors can have better returns if they diversify their product portfolio (Table 14). DP1 profits may increase by 65% after reducing 45% the milk used to produce fresh cheese by producing more aged cheeses, yogurt and manjarblanco (a product based on the reduction of milk and sugar). DP2 obtained 60% more profits by replacing 20% of the milk from fresh cheese to produce aged cheese. The two dairy processors also suggested the simulation of processing more milk volume in order to do not affect their level of fresh cheese manufactured. DP1 increased their profits in 45% and DP2 in almost 100% if they collect one third more of milk. Nevertheless, competition for milk supply in the area would make this second alternative difficult to implement.
126.96.36.199. Simulation of quality-based payment systems Based on the evaluation of milk chemical composition per farmer and on a 11.6% total solids baseline defined jointly with DP1 and DP2, up to 70 percent of dairy farmers receive lower milk prices than they should receive and around 25-30% of them are overpaid. Although the lack of milk chemical quality control seems to be advantageous for dairy processors, this situation is quite risky because under a context of high competition for milk supply they can be left for other processors who offer higher milk prices. A simulation was also conducted to estimate the economic impact of the application a milk
system considering an increase of 0.3% of total solids after the implementation of a bonus of 0.03 per unit above 11.6% and a penalty of 0.10 per unit below showed differences on dairy processor total profits of less than 5%. The present analysis did not show higher differences in terms of profits mainly because we used as a baseline the percentage of total solids recommended at national level and not the average values in the study area. Indeed, no main effect was observed because the two smallscale dairy processors analyzed receive already milk with higher level of total solids. Nevertheless, the simulation gave to small-scale dairy processors the possibility to better estimate the maximum amount of money they can pay per liter of milk to each of their dairy farmers. Moreover, dairy processors realized that if they apply a quality-based payment system they will be reducing the overpaid farmers, rewarding correctly farmers who provide good quality and ensuring suppliers’ loyalty without increasing considerably their milk cost.
4.3. Discussion Although the use of simulation models are essential in dairy industrialized countries to evaluate “exante” potential solutions to given issues such as selecting a milk price or dairy product portfolio and the potential impacts of manufacturing processes on their performances (Geary et al., 2010; Roupas, 2008), there is few literature available regarding simulation tools adapted to small-scale dairy processors or considering milk-quality based payment systems in developing countries. This provide originality to our simulation model DairyPlant, even when it can be seen as a simple dairy processing model compared to the sophisticated predictive models for the dairy industry in developed countries.
However, DairyPlant show two main limitations. Firstly, mechanistic relationships between payment system and farmers’ quality changes were not included in the analysis. Indeed, such relationship is difficult to establish in a specific production context, since it is technically uneasy to link feeding or milking practices to a given quality value (Fuentes et al., 2014). However, changing practices needs for the farmer that extra-costs will be compensate by better milk price (Valeeva et al., 2007), which complicates the modeling of such a relationship. Botaro el al. (2013) reported similar constraints regarding changes in milk composition after rewarding dairy producers. (ii) Calculation in the farmer’s
include individual farmer’s production costs because it would assume that either the dairy processor knows this private information, or farmers agree to give it in a negotiation process with the processor.
Since information on individual production costs is a strategic resource both on farm and dairy sides in such a negotiation, it seems unnecessary to integrate it in DairyPlant.
The first trial of DairyPlant received a positive reaction from small-scale dairy processors in the studied case. Indeed, managing milk manufacturing processes and planning milk incentives systems were unknown concepts by stakeholders when this study started. Simulating different scenarios including different dairy portfolios and milk quality payment systems helped small-scale processors to clarify their ideas about these concepts. Most processors realized the need to control milk quality, since it has a direct effect on their performances and economic revenues. Nevertheless, they also stated that stricter controls could push milk suppliers towards processors who are less interested in quality aspects. In such a context, the implementation of simple quality-based payment systems that guarantee win-win scenarios for all the stakeholders could be a key element.