3. JUSTIFICACION
3.4. EL IMPACTO DE LOS DESASTRES
The Cost Reports module has been implemented partially. It is possible to select products in a subject list for which a cost report has to be generated (Figure 6.28 and Figure 6.29). For every subject list, multiple cost reports can be defined (Figure 6.28). For every cost report, all objects from the three information structures and the cost structures can be selected (Figure 6.30). The information in a cost report can be arranged per subject or per subject list. A cost report can contain information about cost estimates and/or information about the actual costs.
Figure 6.28 Global definition of cost reports.
Figure 6.30 Definition of the information of the subject list to be displayed in a cost report.
6.4 Variant based cost estimation
The variant based cost estimation algorithm was implemented as a DLL as well, so it can be used by any engineering task. Also, a separate executable for the variant based cost algorithm was made to allow for the manual use of the algorithm. The algorithm uses three databases: one metabase, which refers to the databases that contain the characteristics and historic product information to be used in the similarity calculation. In the configuration overview (Figure 6.31), historic products can be selected and categories for the similarity calculation can be defined. For every category it can be indicated whether the category has to be used in an automatic similarity calculation (Figure 6.32). If a category has to be used in an automatic calculation, the desired percentage of similarity has to be indicated as well. For every category, the category items and their determinative quantities can be defined (Figure 6.33).
Figure 6.31 The configuration of the variant based cost estimation algorithm.
Figure 6.32 Definition of categories.
The user interface for manual use is shown in Figure 6.34. Here, by means of defined categories and category items, a product can be described. The categories and category items can be selected from the defined categories and category items (Figure 6.35). Furthermore, the number of category items, the value of the determinative items can be set and the desired percentage of similarity can be indicated. The type of product has to be indicated as well. Based on the input defined, a percentage of similarity is calculated for every product in the “historic database”. When a product meets the desired percentages of similarity, the product is shown in the output field together with its costs and the percentage of similarity per category (Figure 6.34). The estimated cost equals the average costs of all the products shown in the output field.
Figure 6.33 Definition of the determinative quantity per category item.
Figure 6.35 The definition of a characteristic.
In case of an automatic similarity calculation, the product information structure for which a cost estimate has to be generated has to be indicated. From this product structure, the categories and category items, which have to participate in the automatic calculation, are collected. The number of category items and the value of the determinative items are collected as well. The same procedure is carried out for every product in the “historic database”. For every historic product, the percentage of similarity is calculated. When the percentage of similarity of a historic product meets the percentages of similarity set for an automatic calculation, the product is used to calculate a cost estimate.
6.5 Concluding remarks
A test of the prototype implementation of the cost estimation system is reported in the next chapter by means of an example. Because not all modules of the cost estimation architecture have been implemented and because the implemented modules have not been implemented completely, not all features of the cost estimation architecture could be tested.
The fact that the modules Data Analysis, Risk Analysis and Data Tuning have not been implemented, causes that many features of the cost estimation architecture are not available in the prototype implementation. Because the Data Analysis module is not present, it is not possible to deduce cost functions from historic data within the system. This limitation can be resolved by using a separate analysis program. However, because of the lack of historic data no cost functions have been derived. It is also not possible to compare cost estimates with actual costs. The absence of the Risk Analysis module makes it impossible to generate accuracy information about the cost functions and to use this information in cost calculations. Because the Data Tuning module is not available, it is not possible to consider different currencies or inflation.
The Cost Model module allows the definition of multiple cost models. The Cost Determination module can only generate a cost estimate for a component when one specific cost model is selected. Therefore, a hybrid cost calculation is not yet possible. It is also not possible to generate a cost estimate based on a pre-set accuracy level or based on the information available. Because the module Cost Reports has only been implemented partially, it is not possible to generate cost reports yet.
Only the described systems for design and process planning are capable of using the information structures. This means that only cost support of the designer and process planner can be considered. Cost control can only be tested partially because the viewing capabilities of the results of a cost calculation are limited.
The implementation of the variant based cost estimation algorithm has proven to be very slow. The search algorithms used have to be optimised in order to be able to use the variant based cost algorithm in a real environment.