'surprise' orders or missed deliveries; (iii) long-term and costly customer care and repair ... In a Small and Medium Enterprise (SME), there are wide varieties of.
Developing a Manufacturing Strategy Based Upon ABC Analysis of Products P. Tomar, G.B. Neighbour, and S. Parkinson The University of Hull
ABSTRACT This paper proposes a methodology for processing product lines within a supplier in the offhighway and niche vehicle industry and focuses upon categorising product lines depending upon their ranking of volume and value. The methodology presents a logical procedure to categorise parts into three “broad categories” analogous to the classical ABC stock control theory. On completion of the categorisation process, product lines are processed by a set of practical rules within a context of an “expert system”. These rules are defined in order to keep minimum inventory, controlled and stable production plan smoothed over the production period, low throughput time and higher delivery performance with the ultimate goal of 100% on-time delivery for the customer. The analysis has showed that there are five classes of product lines (A1, A2, A3, B1 & B2) which vary in their requirements for planning and operations as outlined in this paper. Evidence is also presented which shows that an empirical relationship, a new “merit” parameter can be determined by fitting a response surface to volume and value quantities and this relationship holds for a wide range of organisations. Indeed the technique can indicate which product lines are behaving in the market as expected and which are asynchronous with the market demand. 1
INTRODUCTION
The analysis presented here is based upon a company manufacturing products for the offhighway and niche vehicle industry (a similar analysis has also been undertaken in the paper products industry which shows very similar results). The constantly changing business environment and an ever-increasing customer demand to improve QCD (Quality Cost Delivery) had the unintended consequences to create erratic delivery schedules and this led to serious balance and flow problems, with excessive inventories in subassemblies and/or raw
materials and work-in-process being held up because of bottlenecks as well as poor information flow. Part of the problem included poor forecast information from the customer which ultimately caused scheduling and re-scheduling almost on a daily basis. These effects increase the level of operating expenses in addition to excessive work-in-process (WIP) and WIP queues. After an initial systems analysis, the problem situation showed that ineffective planning of production occurred due to the “fire fighting” of the production team. It goes without saying that without effective change the eventual effects will be: (i) losing business to competitors; (ii) high investment in overtime and premium manufacturing costs to ship ‘surprise’ orders or missed deliveries; (iii) long-term and costly customer care and repair practices; (iv) increasing safety stock inventories; (v) waste in manufacturing capacity and resources; and (vi) missing opportunities for market turn for future business and missing financial goals. 2
PROBLEM DEFINITION
Customer requirements in terms of logistical performance have become more and more exacting. A key objective of the organisation in question is to achieve 100% on-time delivery performance; excellent customer service is a high priority. This objective has led to the decision that the analysis on volume (or more specifically the management of volume) has a higher priority (& is the primary determinant) from an operations perspective than value (at this time). This is not to say value is ignored, but rather dealt with below volume from hierarchical view. It is also interesting to note that elsewhere the typical ABC analysis is commonly on the basis of value, but the two variables can lead to very different results as shown in Figure 1. The results show that a classical ABC Pareto analysis comply with expectations for either value or volume against (ranked) lines, i.e. the top 20% of lines accounts for approximately 80% of volume / value. Data Sorted By Volume 600
1
Product Lines
0.8 0.7
400
0.6 300
0.5 0.4
200
0.3 0.2
100
Lines Volume
0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Frac. Cumulative Volume
0.9 500
0.1 0 1
Frac. Cumulative Value
FIGURE 1: The relationship between product lines and cumulative volume and value.
Traditional “dollar usage” classification has been criticised by many researchers because it neglects many factors. In a Small and Medium Enterprise (SME), there are wide varieties of variables influencing inputs and outputs that affect the way the organisation performs. Identifying these variables and optimising them require prioritising the key business requirements, but what are these? These key business requirements could be based on quality, delivery, price, branding and so on. A typical SME will have a moderate product mix, but in a high variety production environment, the situation becomes very complex. It, therefore, becomes paramount to process these products through the manufacturing stream with a well defined methodology based upon key business requirements. Here, after following an ab initio approach, the analysis drew the authors to a variation of the classical ABC Pareto approach. The principal aim being to streamline the production process within the manufacturing system to reduce the influence of erratic customer demand typical of that found in JIT systems as well other issues often interpreted as ‘fire fighting’. There is a school of thought that rather than concentrate on either volume or value, a pair-wise matrix of volume / value rankings can be created in which 9 classes are obtained such as AA, AB, AC, etc.. However, this implies volume is of equal weighting to value. From an analytical view point, it is intuitive to think that a “merit” rating for each line can be generated which forms part of a response surface (z-axis) on the plane of volume (x-axis) and value (yaxis). There is nothing in the open literature which suggests the form of the response surface (i.e. fitting equation) and indeed this appears to be a novel idea. It should be mentioned that from a dynamic perspective, others have developed complex mathematical models to inform decision making such as the Wagner-Whitin1 algorithm. Indeed, more complex multiple criteria ABC analysis, based upon analytic hierarchy process (AHP), have been undertaken by others2 to generate a “criticality index” which includes consideration of a range of different factors, e.g. lead time, unit price, durability, order size requirement, etc. For example, Flores and Whybark3 have proposed a similar bi-criteria approach and used standard ABC analysis on each of two criteria independently and then combined the two single-criteria ABC grouping through the use of a matrix. In terms of applicability to our analysis here, this list could include part size, stillages available, capacity of a stillage, etc. The number of criteria or alternatives should be reasonably small to allow consistent pair-wise comparisons. Satty4 suggests a maximum number of seven. ..The implication here is that both low and high value or volume lines can have high (or low) criticality. The criticality index has some equivalence to the merit rating proposed above. Given that this is accepted then there remains some ambiguity for the relationship between merit and the xy-plane. Various models have been investigated and fitted to a response surface with limited success. One of the more successful models, illustrated below, with R2>0.9 is:
Merit =
Value − Valuemin Valuemax − Valuemin
Volume − Volumemin Volumemax − Volumemin
Equation [1]
The evidence presented here shows that an empirical relationship of a new “merit” parameter can be determined by fitting a response surface to volume and value quantities and this relationship holds for a wide range of organisations. Indeed the technique can indicate which product lines are behaving in the market as expected and which are asynchronous with the market demand5. The analysis has shown five classes of Product lines (which the authors
have denoted A1, A2, A3, B1 & B2) which vary in their requirements for planning and operations. Aspects of the methodology include: •
• •
Standardizing the length of the works order interval (cyclical planning) aiming at short throughput times through synchronous manufacturing (high service level & low inventory), reducing WIP and co-ordination of the supply chain (A1 & A2 lines). Synchronizing the ordering of a batch of products with the reordering of exactly the amount of required parts and raw materials (applicable to low cost B1 lines). The management of low-value items to avoid stock-outs and to reduce costs such as procurement and material handling (A3& B2 lines).
FIGURE 2: The Response Surface generated from Equation [1]
3
PROPOSED SOLUTION
As mentioned previously, the most important Key Performance Indicator (KPI) for the company in addition to reduction in inventory and lead time is ‘Delivery Performance’. The adopted methodology has to differentiate between product lines to ensure optimum lead time and inventory. Based upon the ABC analysis, information in the new methodology is used to prioritise the processing of parts through manufacturing system. Demand from customer for finished product lines is the starting point for the methodology. Data for finished lines from last ‘12’ months is collected. Pareto analysis on volume indicates that the first 20% are categorised as A and the remaining 80% are classified as B. As discussed, product lines are further sub-categorised as A1, A2, A3, B1 and B2. These product lines are then further
processed based on a set of rules as reflected in Figure 3. The in-depth analysis has shown, perhaps surprising, support (& validation) from the open literature which is not widespread, but what is present is robust6. It must be stressed that the philosophy, principles and practices reported have been somewhat developed independently from a logical and methodical approach with the specific application to the company. Therefore, some comfort and confidence can be held in the conclusions reached so far. It is interesting to note that some papers in the open literature have concluded that there really only exists two classes of lines, A and C, whilst others have concluded, on the basis of lead time requirements, that there are in fact five classes of items (as is the case in this analysis). The new system, giving priority to volume (through demand), should reduce WIP and thus mean shorter lead times and greater flexibility. The new system is currently being monitored using the following measures of performance: (i) Delivery Schedule Achievement (and non compliance with time / quantity); (ii) raw materials inventory expressed in days in comparison with the rate of sales; (iii) WIP as a measure of investment; and (iv) lead time reduction.
FIGURE 3: Methodology of processing parts through manufacturing system
4
CONCLUSIONS
Today' s marketplace requires high velocity supply chain performance. On-time delivery and customer service go hand-in-hand. Customers can, and will, change suppliers when they are unable to get the goods they want in the response time they need them. The objective is to have the ability to provide customers with what they want in exact quantities they require7. The methodology presented here has the potential to streamline the planning process and production processing system with the help of the proposed “merit” rating. This will improve KPI’s in terms of delivery performance, lead-time and inventory levels. The grouping will provide management with more effective means for specifying, monitoring, and controlling system performance, since strategy objectives and organisational factors can often be represented more naturally in terms of strategic groups.
REFERENCES 1 Wagner, H. M. and Whitin, T. M. (2004). Dynamic version of the economic lot size model, Management Science, Vol. 50, [12], 1770-1774 (first published 1958). 2 Waters C. D. J. (2003). Inventory control and management (2nd Ed), Wiley, Chichester. 3 Satty, T.L., (1980). The analytic hierarchy process, McGraw-Hill, New York. 4 Partovi, F. Y. and Burton, J. (1993). Using the analytical hierarchy process for ABC analysis. Int. J. Operation & Production Management, Vol. 13, [9], 24-44. 5 De Smet, R. and Gelders, L. (1998). Using simulation to evaluate the introduction of a Kanban subsystem within an MRP-controlled manufacturing environment. Int. J. Production Economics, Vol. 56-57, pp. 111-122. 6 Bicheno, J., Holweg, M. and Niessmann J. (2001). Constraint batch sizing in a lean environment. Int. J. Production Economics, Vol. 73, [1], 41-49. 7 Christopher, M. (2005). Competing through supply chain excellence - presentation notes from Supply Chain Management Presentation at the University of Hull.