Autores: José Antonio Heredia Álvaro, Ignacio Monserrat Soriano , Antonio Estruch Ivars , Nico Vandaele, ESID Universitat Jaume I, Castellón, Spain, Industrias Ochoa, Valencia, Spain, KU Leuven, Belgium
Estudios previos sobre la frecuencia de actualización de los parámetros del plan de producción dentro del contexto de MRP.
Esta investigación contribuye al resto de estudios, mostrando como un SME puede mejorar la efectividad mediante unos Planes de Control de la Producción (PPC) más adaptados y aplicados a la base teórica.
En este estudio se describe como el personal de PPC con la colaboración activa de investigadores usan un DSS predictivo basado en una simulación para actualizar los parámetros del plan de producción y como estos cambian los mecanismos de coordinación tanto a nivel táctico como a nivel operacional.
Dado de que se trata de un proyecto de investigación limitado a un único caso de estudio, no se pretende generar conocimiento universal con este estudio, sin embargo, se discuten sobre diferentes características clave que pueden ayudar a explicar la agilidad de la empresa para adoptar las nuevas prácticas; y esto llevara a más investigación en el futuro.
Es un artículo muy amplio que lo vamos a ir mostrando en diferentes posts a lo largo de estos días para poder leerlo tranquilamente POR CAPÍTULOS
CAPÍTULO 1
Purpose and rational of the research
The production planning and control (PPC) process, trying to balance supply and demand, affects most of the manufacturing costs and service aspects offered to the customers significantly.
It is well known from literature that the performance of the PPC process is determined by the planning decision variables (i.e. lot size, safety stock, planned lead-time, planning periods, available capacity) utilized to decide for each product: when to produce, in which quantities and on which resources.
ERP/MRP systems, are standard software packages supporting the PPC process that need to be configured and parameterized (Fransoo and Wiers, 2008; Stevenson, et al. 2005).
After decades where the focus was on implementing ERP/MRP systems for production planning, now we argue that companies need to learn how to update its planning decision variables dynamically. Internal changes as a consequence of shop floor process improvements, new products offerings, updated company targets, are examples of changes that demand the review of the current value of the planning variables. Externally, many industrial environments are characterized by rapid changes in financial conditions, product life cycles, customer preferences, competitors’ offerings, and technology developments that logically also should affect the adequate value of the PPC decision variables. Maintaining the PPC process with fixed decision variables values, that disregard these internal and external changes, may lead to severe performance deterioration (Vandaele and De Boeck [2003], Van Nieuwenhuyse et al. [2011]).
Therefore, we understand that companies should establish some kind of internal organization with the purpose to keep the planning decision variables values (which from now on we call parameters) synchronized with reality. This organization should be especially relevant for companies looking for operational excellence.
Due to the difficulties in finding business cases with a formalized organization for updating its PPC process, we decided to set up an action research project to better understand the entire problem complexity and the pitfalls that companies may experience when trying to implement this new practice. The selected company is seen as representative of SME in the metal working sector.
The two first paper´s authors, an Operations Management professor and the Operations Manager of the company, engaged in a long term relationship, agreed to sign a contract between the company and the University to perform the research activities.
Subsequently, a group of assistant researchers was set-up and weekly meetings were scheduled at the company premises with operations and production personnel. The third author was focused on developing the information system to support decision making. The development of a simulation-based model as a core component of the decision support system (DSS) guided the data gathering and performance analysis process. The DSS provided a “virtual planning environment” suitable to assess the current operations capability and to test the impact of different planning methods and parameters on customer service and cost performance measures.
Company description
The company selected, Industrias Ochoa, is a SME manufacturer, characterized by producing multitude of items, limited but flexible production resources, and with stochastic and lumpy demand. In this scenario, the different planners are autonomous decision-makers but as we will show in this paper, the performance of the overall system rely on how well they coordinate their decision-making about the planning parameters.
Industrias Ochoa is located in Valencia (Spain) with 250 employees, and manufactures more than two thousand finished products, more than nine thousand semi-finished products based on more than 300 components and more than 300 raw materials. In house Manufacturing processes include stamping, machining and welding. Manual assemblies are outsourced to neighboring installations. The facility layout is mainly organized in a process production typology although there are several product oriented production manufacturing cells where product families are produced as in a ‘factories within a factory’ mode [Milterburg, 2008]. The production is organized in one or two shifts coordinating the different processes.
Industrias Ochoa (from now on IO) participates in two major supply chains. The company is part of the Ford supply chain as Tier 2 and Tier 3 supplier depending on the product and also it is part of the Hilti supply chain for different types of small size parts.
The operations department staff develops the production planning using a MRP system, and the schedulers, in the production department, realize the detailed scheduling.
The MRP system is linked with the plant data sources. Historic and actual plant, planning and warehouse up-to-date data are available.
We selected a simple but representative product family within the Hilti supply chain. It is a group of 20 clamps articles used to support pipes in construction projects (e.g. buildings, large ships, industrial facilities). Its annual production is around 1.5 million units. The products are sent to Hilti stores (inbound logistics) or to warehouses for distribution (outbound logistics). Best service is the key objective for this product. This means that the production process priorities are: production flexibility; quick response in terms of orders processing and delivery; and to engage the necessary stock to serve the orders within the due date.
For most of the articles studied within this family, the demand throughout the year is quite jumpy, with occasionally larger orders than usual intended for special building projects. There is a Service Level Agreement with Hilti to provide the orders within the agreed lead time of 6 laborable days with a service level of 95%.
The details of the agreement are renewed each year. An annual meeting is held in the Hilti headquarters to review the performance and establish the guidelines for the next year. Hilti provides each month a forecast update for the next four months. The agreement states that the average forecast error has to be less than 50%. To obtain the monthly forecasts, Hilti applies a technique based on the Croston´s method [Croston, 1972], as included in the APO module of SAP, over the past monthly data using 36 months. Then, the monthly demand forecast is distributed uniformly between weeks as input to the MRP.. During the period of the study (2012-2016) the demand followed a growing trend. We checked that during this period real demand exceeded the forecasts systematically. We observed that in many occasions the forecast error was clearly more than 50%.
Taking into account this level of uncertainty, the finished product warehouse (outsourced to the logistic company DHL) maintains high stock levels.
The production planning and control (PPC) process is organized at three levels. The operations manager, with more than 30 years of experience, analyses the forecast requirements to identify possible mismatches between capacity and demand for the mid-term tactical resource planning.
As a consequence some manual changes in the forecast input to the MRP are introduced trying to level supply capacity with demand. The Operations department staff (second level) manages the MRP system to generate the Master Production Schedule (MPS). The different product families are divided between the four permanent staff that adjusts manually the dates and quantities of the production plan orders. They also expend a considerable amount of time (around two complete working days) modifying the previous plans to reflect the actual executed schedule. The schedulers (third level), assigned to the production department, perform the detailed sequencing and scheduling. They apply an ad-hoc mental scheduling heuristic synthesized from past experience to sequence the manufacturing orders included in the MPS. A running project since 2011 to implement ASPROVA software for scheduling has not resulted in an effective use, because the actual manual procedure consistently outperformed the simulated parallel outcomes obtained by the software.
Action research objectives
In this action research, the immediate objective for the manager was to increase the company analytical capabilities in order to help to reduce the inventory costs maintaining the high service standards. From an academic perspective the initial purpose of this research was to learn about the dynamics and the details involved in rolling horizon planning and scheduling and to identify ways to coordinate better the dynamics of the planning process in which multiple planners are involved. To support the research, both parties agreed on the development of a decision support system, modular and scalable, which could easily be extended to the remaining products families. For analysis purposes, a hybrid discrete event and agent based simulation model using the software Anylogic and its Java embedded language has been developed. The simulation model includes a model of the planning logic and a model of the product´s manufacturing supply chain. This software also includes optimization engines that may be used to perform experiments in order to find adequate values for the planning decision variables, (i.e. the MRP parameters). The process of designing, developing and validating the software environment is explained in detail elsewhere.
Structure of the paper
In the next section we provide a background of the research question by reviewing the literature related to the dynamics of production planning using a MRP system. Then, the remainder of the paper is organized following the action research guidelines provided by [Coughlan & Coghlan, 2002], as follows: methodology review and method of inquiry, data gathering, data feedback and analysis, action planning, evaluation and outcome. In the section of data gathering we justify why we developed a Decision Support System based on simulation and how this development guided us through the data gathering process. After analyzing the actual planning organization, we realized that it was necessary, not only to compute better the production planning parameters but, also to change the coordination mechanisms between the different planners involved.
The action planning section describes how a working group is set up to review systematically the planning methods and to update the values in the information systems supporting the PPC.
Finally, in the evaluation and outcome section we reflect about the different aspect learned and provide the conclusions of the research, including implications for further research.
Autores: José Antonio Heredia Álvaro, Ignacio Monserrat Soriano , Antonio Estruch Ivars , Nico Vandaele, ESID Universitat Jaume I, Castellón, Spain, Industrias Ochoa, Valencia, Spain, KU Leuven, Belgium
Próximo CAPÍTULO
CAPÍTULO 2
Background and state of the art
