Spare parts demand forecast in maintenance management
Summary
The availability of spare parts at the time they are needed for replacement in the event of a failure is an essential element to achieve adequate compliance with the maintenance plan established in the company. In this article, the authors present a procedure that allows forecasting the future demand for spare parts within production equipment, based on their reliability.
Keywords: Maintenance Management, Demand Forecasting, Spare Parts
1. Introduction
Experts are always trying to make better estimates about what will happen in the future when faced with uncertainty. The primary purpose of forecasting is to make good estimates on which to base decision-making models. Forecasts are the fundamental problem within the management of a company's activity due to the complexity of the problems encountered when forecasting and their impact on all company decisions.
A very particular situation presents the forecast of the demand for spare parts, given the large number of assortments that must be provided, the high degree of variability in the demand for these productions and the importance of the availability of these parts for the equipment. that uses them, since their availability in time guarantees the continuity of the production process in which the equipment that uses them is involved within a certain company. That is why a high level of precision must be ensured in forecasting the demand for this type of product.
This article presents a general procedure for forecasting the demand for spare parts based on their operational reliability based on the philosophy that characterizes the system and considering its influence on the improvement of Equipment Maintenance Management. This general procedure branches out into more specific ones, elaborated by virtue of the main components of any demand forecasting system, which are:
Determination of the current situation.
Collection and analysis of information.
Determination of the most suitable forecasting method.
Monitoring and control of the forecast.
Monitoring of the implementation of the procedure.
Consequently, the content of each one of them raises the steps to be developed to anticipate the demand for spare parts based on their reliability, sometimes defining the specific techniques to be used.
2. Definitions and general concepts
Below is a set of definitions and general concepts that are considered necessary for a proper understanding of the elements to be exposed within the proposed procedure.
- Forecasting (1): it is a process in which data is collected and analyzed to estimate what will happen in the future with a certain factor in an uncertain environment. This concept raises three main areas of attention: (I) the forecast period, (II) the specific variable to be forecast, and (III) the forecasting technique to be used.
There are different types of formal models for forecasting demand. These have been divided into two large groups for study, one is the quantitative analysis and the other is qualitative access. Quantitative forecasts handle a variety of mathematical models that use historical data (time series) and / or causal variables to forecast demand. Qualitative or subjective forecasts incorporate important factors such as intuition, emotions, personal experiences of the decision maker, and value systems to reach a forecast. An overview of the different types of forecasting methods available is shown in Figure 1.
(1) Some authors use the words prediction, forecast and forecast interchangeably.
Figure 1. Overview of forecasting methods
Reliability: is the probability that an item (system or element) satisfactorily performs the specified mission, during a given period and under a given set of operating conditions.
- Failure patterns:
For decades, conventional wisdom suggested that the best way to optimize the performance of physical assets was to restore or replenish them at fixed intervals. This was based on the premise that there is a direct correlation between the amount of time (number of cycles) that the equipment is in service, and the probability that it will fail, as shown in Figure 2. This suggests that the expectation is that the Most items will operate reliably for an “X” period, and then wear out.
Figure 2
Indeed, this predictable correlation between "age" and "failure" is valid for some failure modes. However, in general the teams are much more complex even than they were a few years ago. This has led to astonishing changes in equipment failure patterns, as shown in Figure 3. The graphs show the conditional probability of failure as a function of age of operation for a wide variety of electrical and mechanical items.
Figure 3.
Pattern A is the well known "tub curve" and pattern B is the same as in Figure 2. Pattern C shows a slowly increasing probability of failure, without a specific age of wear. Pattern D shows a low initial probability and then a rapid increase to a constant level, while Pattern E shows a constant probability at any age. Pattern F begins with a high probability of infant mortality to decline to a low and constant or slightly increasing probability of failure.
- Failure: event from which the article ceases to fully or partially fulfill its functions within defined limits of action. This is nothing more than the cessation of the article's work capacity status.
It should be noted that in calculating reliability, one of the greatest difficulties is faced with the insufficient database on the occurrence of failures, mainly due to the absence of a correct organization to collect them. Frequently, this is compounded by objective difficulties related to the complexity of collecting information during exploitation.
3. Procedure for the improvement of the forecast of the demand for spare parts
To forecast the demand for an item (spare part) based on its reliability, during the exploitation stage, it is necessary to define a broad procedure made up of several interrelated stages, which must allow a complete and exact evaluation of the reliability indices that provide relevant data for the purpose of this research and, based on said evaluation, define the forecasting technique to be used. Figure 4 represents the proposed procedure.
Figure 4 Demand forecast improvement process
The procedure begins with the definition of the philosophy, which must be the focus of the system since it constitutes the policy that will permanently govern its performance.
Then, the determination of the current situation is carried out in order to define, based on the analysis of a series of indicators, the characteristics that the demand forecasting system presents at that moment, for later, based on said situation, to propose some improvements to make the forecasting process more efficient. Once the improvements have been proposed, it must be verified whether they achieve the desired and necessary levels of precision and propose new improvements if the previous ones are not enough or move on to applying them otherwise. The application must be linked to constant monitoring that provides feedback to the system, to take the necessary measures in the event of any disturbance.
Each of these four stages will be explained in greater detail in the following sections.
3.1. Definition of the Philosophy of Demand Forecasting
The first stage of the procedure is precisely the definition of the philosophy of the system, since this allows to conceive a set of doctrines, challenges and principles that are in accordance with the conditions that emanate from the environment, the strategies and the objectives set by the company. Based on this, the philosophy of the system is defined as follows:
Achieve an increasingly accurate forecast of the demand for spare parts, based on the performance of the reliability of the equipment for which they are intended, by adopting the forecasting method that best and fastest adjusts to the behavior of the same.
In addition, the philosophy needs to be based on a series of well-defined objectives that flow throughout the organization and are translated into measurable terms, which allow the daily work of the company to be carried out and have adequate criteria for the most efficient decision-making. and effective as possible. These objectives are as follows:
Guarantee the required accuracy in demand forecasts.
Achieve rapid adaptation to sudden changes in demand.
Reduce the downtime of productive equipment, especially limitations.
Reduce uncertainty in decision-making around future demand.
Reduce the number of unforeseen events or disturbances in production plans.
In applying these goals, it is necessary to recognize that not all can be achieved with the same degree of success. However, it should be noted that the definition of the objectives is not necessarily the same for all objects, since it depends on the conditions present in each specific situation (company) in which the work is carried out.
3.2. Determination of the Current Situation of the System
This step is of great importance, since it consists of carrying out two parallel analyzes (internal and external) that will allow, together, an evaluation of the current and potential situation of the demand forecasting system used, that is, it allows find out what the main characteristics and problems of said system are. The analysis of the results of this stage allows greater attention to be paid to the variables that most affect the achievement of the proposed objectives and to those that in one way or another were neglected up to now.
3.3 Definition of System Improvements
Knowing the desired state and the one that the system really presents, it is unavoidable to look for improvements that allow us to advance from one to the other (overcome the gap between them). These improvements can include from small modifications in the system to the redesign of the same according to the particularities that arise in each case.
To define the system improvements, several actions must be carried out (see figure 5): (A) selection and approach of the forecasting method to be used, (B) realization and refinement of the forecast, (C) monitoring and control of the forecast and, (D) presentation of forecast results. A detailed explanation of each of these actions is provided below.
Figure 5. Procedure for defining system improvements
A. Selection and approach to the forecasting method to be used
It is in this step that the influence of the results of the article's reliability studies on the demand forecasting procedure is directly observed.
As could be seen in the bibliography, there are a series of factors that influence when deciding the forecasting technique to use; However, taking into account the main objective of this work, the fundamental role falls on the stage of the life cycle (from the point of view of its exploitation) in which the article is found. In addition, this is complemented by the generalized criterion presented by a considerable number of authors about the need to base forecasts of the demand for spare parts on analysis of their reliability, although it has not yet been developed. a procedure that makes these approaches a reality.
Figure 6 shows the different forecasting techniques that, in the opinion of the authors, can be used to forecast the future demand of the article based on its reliability, recommending to use, first, those simpler methods, as long as not extremely accurate forecasts are required.
Figure 6. Life Cycle Stage Relationship - Forecasting Methods
In the case of the other failure patterns, an analogy is made between the methods used in the different stages of the "bathtub curve" and those that can be used when the article follows some of these patterns. The analogy would be established as follows:
When the behavior of the faults tends to grow, the forecasting methods proposed for stage III of the bathtub curve are used.
When the behavior of the failures is stable, the forecasting methods proposed for stage II of the bathtub curve are used.
When the behavior of the failures tends to decrease, the forecasting methods proposed for stage I of the bathtub curve are used.
The value of the parameters of the forecasting model to be used is decided by a trial and error mechanism. The process can be simplified by using existing software for this purpose.
B. Refinement of the forecasting process
This action aims, firstly, to carry out an analysis of the forecasting methods proposed to be used at each stage of the product's life cycle (figure 6) during its exploitation, to decide which of them are true modifiers of the real demand for the spare parts by means of a Multiple Correlation and Regression analysis (see equation 1) and, to form the integration equation with a view to considering the advantages of those more precise methods when making the forecast; to later decide which interval of values is more advisable to assign to each of the weights (bj) that modify these methods in the integration equation (see equation 2), intervals that will help in making decisions for future work on this matter.
C. Monitoring and controlling the forecast
The monitoring and control of the forecast includes the supervision and control of the forecasts made to ensure that they are being carried out properly. Currently, the most used way to monitor forecasts is the use of a Tracking Signal, which must be kept within Tracking Control Limits for the selected forecast model to follow. being valid. In this way, when the tracking signal exceeds the control limits, the forecasting process must be stopped and demand re-absorbed and matched more accurately (model correction and / or forecast method).
D. Presentation of forecast results
To present the forecast results, we start from the knowledge of the probability density function that follows the historical demand. Then, considering that the forecast must be an undivided projection of what is expected to happen, the forecasts will be based on probabilities. By basing forecasts on probabilities, contingencies can be planned and the risk inherent in each decision can be properly appreciated. If these estimates are not available, the plan cannot adequately anticipate contingencies or, worse, will be set at mediocre or minimal levels so that it can be exceeded.
3.4 Implementation and monitoring
Once the improvement that best approaches the desired and necessary levels of precision has been decided, we proceed to its practical application. The application must be linked to a constant monitoring that provides feedback to the system, to take the necessary measures in the event of any disturbance (unwanted variations with respect to what is expected).
The company's methods must be evaluated against the characteristics of the “state of the art” system; that is why the final key to success is follow-up during its application. It is necessary to periodically audit the results achieved in relation to those projected, monitor the obstacles to success, and reestablish the guidelines.
During the monitoring of the behavior of the system, in the present work, it plays a fundamental role to determine the existence of errors during the forecasting, which can be obtained by executing a forecasting test.
This is a simple and very important step that gives a measure of how useful the forecast has been for the company. This step, in general, consists of determining, after a time similar to the Mean Time Between Failures (TMEF) or the Mean Time Until Failure (TMHF) calculated for the article, the status of the improvement carriers to check if the Proposed improvement is yielding the results for which it has been designed.
This constitutes the main carrier of the improvement of the system since it is the one that guarantees that the downtime of the machines is kept to a minimum. The situation in which the improvement bearers find themselves at the end of each Mean Time Between Failures will be fed back to the step corresponding to the analysis of the current situation of the system in order to, in case of disturbances, make the necessary modifications to achieve the objectives paths.
3.5 Advantages of the proposed procedure
The result achieved with the procedure has a great impact and brings benefits for other key areas of the company such as:
Decrease in losses due to productions stopped due to prolonged stays of the production equipment.
Decrease in the amount of unforeseen events in the development of maintenance programs.
Establishment of a more efficient spare parts inventory policy.
Improvement of the balance in the utilization of the maintenance workforce.
Increased level of customer service.
4. Conclusions
Once the work is completed, the following conclusions are generally reached:
- The nature of the products under study, where the demand depends on the moment in which the failure of the part that is installed in the production equipment occurs, determines the importance of conducting reliability studies in order to guarantee that at the time of failure Even though the testing phase is the last step in the general procedure proposed for forecasting the demand for spare parts, it must have a permanent character of adjustment and improvement, since only in this way can the process be achieved of continuous improvement that is necessary for this type of production In these moments, when demand patterns are constantly changing, the proposed procedure for forecasting the demand for spare parts,constitutes a valuable tool that contributes to the achievement of the competitive level that is required of all types of organizations.
5. Bibliography
- Anderson, DR et al "Introduction to quantitative models for administration" Grupo Editorial Iberoamericana. 1995.Creus Solé, A "Reliability and safety of industrial processes" Edit. Marcombo, SA Spain 1991. Fernández Sánchez, E and CJ Vazquez Ordás “Direction of production. Operating methods ”2nd Part. Edit. CIVITAS, SA 1st edition. Spain. 1994.Goldratt, EM “The goal. A process of continuous improvement ”Ed. Jaular. Spain. 1990/1 ------ “The race” Ed. Jaular, SA Spain. 1990/2….. “Custom Maintenance Management Manual” published in: Mathur, K and Solow, B “Operations research. The art of decision making ”, Ed. Prentice-Hall Hispanoamericana, SA 1st edition in Spanish, Mexico, 1995. Maynard, HB“ Manual of engineering and industrial organization ”, Ed. McGraw-Hill Book Company, New York, USA, 1990.Moubray, J. "RCM 2:Maintenance strategies, a new paradigm ”, published in: Pèrez Jaramillo, CM“ The future of the Maintenance Function ”. Render, B and Heizer, J“ Principles of operations management ”, Ed. McGraw-Hill Interamericana de México, SA 5th Edic. 1991. Yepes de Castaño, AM and Alvarez Villa, ME "Forecasts using probabilistic models: a tool in decision-making", Rev. Universidad EAFIT, # 106, 1997.1997.1997.
