UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Life time buy modeling at Agilent Technologies Rugina, Dragomirna


The PLM (Product Lifecycle Management) Toolkit at Agilent is a decision-support system used in support of Product Lifecycle Management decisions across the supply chain. The development of the tool started 4 years ago and new features are currently added. Utilization of the tool includes product portfolio decisions, product rollovers and discontinuance, lifetime buy decisions and contract manufacturing. The manufacturing environment at Agilent is high mix, low volume and it poses several challenges. Among the unique challenges that TMO (Test and Measurement Organization) faces are long lifetime of the products and assurance of supply for products whose components have shorter lifetime than the product itself. When a part is discontinued by the supplier, it is necessary to make a life-time buy, that is to buy enough of that part so that it can be used in all the products for their lifetime (dependent demand) and ensure all support (independent demand). We call such a component a lifetime buy (LTB) part and we call the LTB process the process of buying for the lifetime of all products containing that LTB part. In the past few years, the users expressed a need to build a stand-alone lifetime buy tool, which would assist decision makers in their current operations involving excess, allocations, assurance of supply and shortages of LTB parts. The current users (planners - engineers) use the LTB tool for LTB calculations. The current LTB tool uses as input the products and computes the necessary quantities of LTB parts assuming infinite overbooking - that is, existence in unlimited supply of all the necessary LTB parts. This assumption is obviously very optimistic, and there is a need to build a tool allowing limited overbooking. In what follows, we understand by overbooking the allocation of parts to a product after one part has actually run out. The current LTB tool does not allow the users to build side constraints (lower bounds on the quantity of product to be built and upper bounds given by production capacity), therefore there is a need for an LTB tool that includes an optimization engine, allowing the users to input side constraints. Another important concept in the LTB problem is equal runout. Equal runout means that all LTB parts run out in their products at about the same time, or equivalently, demand in immediate periods is preferred to demand in later periods. The purpose of the paper is to analyze and optimize of the lifetime buy process in order to improve the lifetime buy decision. The paper gives the description of the lifetime buy tool that has not being yet implemented. Chapter II of the thesis is dedicated to the determination of a runset on which optimization is performed. In the existent tool, a runset is composed only of a set of products and their components (a single explosion down or up through the bill of materials). For optimization purposes, we propose the utilization of larger runsets, obtained by iterating consecutive explosions up and down. The criteria for stopping the iterations are given by the products in the runset or the LTB parts at the frontier of the runset. Chapter III presents two different one-period optimization models: A linear model for the optimal use of LTB parts on hand, structured as a trade off between maximizing profitability, minimizing deviation from equal runout and maximizing customer service level represented by insurance of parts for support. A linear model for the case of overbookings, structured as a trade off between maximizing profit, minimizing deviation from equal runout and minimizing value of overbookings. Two objective functions are built for the deviation from equal runout, one for the one period case and the other for the multiperiod case. Also, a binary model is built in order to build the number of overbookings objective function. The MIP solver is not available at Agilent, therefore the model containing the binary variables cannot be implemented yet. Heuristics are built for the overbooking of LTB parts as follows: number of overbookings is limited by a constant; value of overbookings in each product cannot exceed a given constant; value of overbookings for an LTB part cannot exceed a given constant. Some of the decisions that can be made using the models are: optimal use of LTB parts on hand, what products are candidates for discontinuance, when a product or LTB part will be discontinued, what LTB parts should be overbooked and in what quantities, how large the redesign investment can be. The legacy system at Agilent is very complex and the management information systems and the MRP databases used are different from site to site. The PLM Toolkit is a management information system that pulls data from the legacy systems at all sites and can be used globally. Given the fact that data necessary for the life time buy model has not being extracted from the legacy system to the date of the presentation of this thesis, in a form that would allow the GAMS program to run, we have not been able to the present date to test and run the GAMS program on a real set of data. Therefore the LTB one period model is applied to a toy data set and run in Excel Solver, in the Example in Chapter III, Optimization. From the modeling point of view, the problem is very difficult because it is clearly a one period, multi-item stochastic model, close to the newsboy problem. However, the demand distributions for the products are not available, and moreover, even if they were made available, it would be very hard to see how demand at the upper level in the bill of materials translated in dependent demand at the raw material - LTB part level. This is the reason why we model the LTB problem using deterministic linear programming.

Item Media

Item Citations and Data


For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

Usage Statistics