Abstract 002-0024

Title A Customer-Orientated Profitability Model for an Australian Timber Company

Conference Second World Conference on POM and 15th Annual POM Conference

Cancun, Mexico

April 30 – May 3, 2004

Authors Peter Palmer

Macquarie Graduate School of Management

Macquarie University, North Ryde NSW 2109, Australia

(E) (T) +612 9496-9167 (F) +612 9496-9170

Willem Selen

Macquarie Graduate School of Management

Macquarie University, North Ryde NSW 2109, Australia

(E) (T) +612 9850-8984 (F) +612 9850-9019

Abstract

In the timber industry, production-orientation is so strongly entrenched that even today the needs of the customer are often overlooked. This paper positions the production setting of the timber industry, with the aim of developing a generic profitability model that may assist in the adoption of a customer-orientated production management approach that is better suited in today’s timber supply chains. The unit of analysis is an Australian timber company engaged in the growing, harvesting, milling, selling and delivery of softwood timber. An efficacy model is suggested, using the input measure of aggregated supply chain cost and an outcome measure of aggregated revenue, avoiding the distraction of transfer pricing and offering a more balanced view across the supply chain.

Introduction

The timber industry is encountering change, just as other industries are. One pressure industries are facing is the trend away from a production-orientation to a customer- (or market-) orientation (Parente, 1998). Customer-orientation focuses the resources of a company on satisfying the needs of a customer, while production-orientation attempts to maximise the utilisation of plant capacity. In the timber industry, production-orientation is so strongly entrenched that even today the needs of the customer are often overlooked.

The consequences of a production-orientation can be observed in the industry. A recent survey of 104 timber industry customers throughout the eastern states of Australia and Tasmania conducted by State Forests of NSW (Tolhurst, Armitstead et al., 2001) revealed a depth of passion directed at producers, critical of perceived inadequacies in supply, branding, pricing, product quality, and pack sizes. “Several respondents were of the opinion that softwood producers are totally out of touch with their market and what makes it tick” (Tolhurst, Armitstead et al., 2001, p. 10).

As part of a preliminary exploratory research of the industry, interviews with a number of senior timber industry executives suggested that the inherently flawed nature of the raw material was the sole driving reason. It is not possible to predict with absolute certainty which products, and in what quantities, will be manufactured from a given stock of logs. Because of this, trying to match production to demand has been fraught with difficulty, and mill management has focused instead on the efficiency of the operation.

This paper positions the production setting within the timber industry with the aim of developing a generic profitability model that may assist in the adoption of a customer-orientated production management approach that is better suited in today’s timber supply chains. A rationale for the production-orientation of the industry is suggested and then explored. The operation of the timber supply chain is described and the difficulties of production planning in this environment are examined in the literature. The literature is also used to examine the link between orientation and profitability leading to a discussion of a supply chain efficacy model. Finally, further work examining the relationship between orientation and profitability in the industry is previewed.

The unit of analysis is an Australian timber company, engaged in the growing, harvesting, milling, post processing, selling and delivery of softwood timber. This company has about 25% of the softwood timber market in Australia, employs over 1000 people, and manages four mills spread along the eastern coast of Australia. The company is currently production-orientated, but there are clear pressures emerging from its customers and its parent company for the adoption of a customer-orientation. Because the issue is topical and relevant within the company and the industry, it is seen as a good candidate for analysis. The production orientation of the timber industry is elaborated on below.

Production-Orientation in the Timber Industry

“Since the early 1980s, in the United States and other developed countries, the dominant force in the seller-customer relationship has shifted. Sellers no longer have the upper hand; customers do. Customers now tell suppliers what they want, when they want it, how they want it, and what they will pay” (Hammer and Champy, 1993, p. 18).

This however is not a widely held view in the timber milling industry. Bryan (1996), in an examination of profitability of a sawmill, rejects this notion: “Organizations allowing the marketplace to define their direction cannot stay aimed at Destination BestPossible” (Bryan, 1996, p. 148). This compact statement very much encapsulates the planning philosophy of the timber mill. The manager of the mill is exhorted to select only those markets and customers for which the mill can be optimised. In this way the weaknesses of the operation are minimised, and profitability of the mill maximised. It is not a large step then to regard the rationale of mill production planning being based on internal factors such as raw material availability and production capability, rather than the order book. The timber mill is identified as a stand-alone business, independent of the supply chain in which it exists, and upon which it depends.

Preliminary exploratory research of the industry suggests that the underlying reason for this production-orientation is that the raw material is inherently and often invisibly flawed, and that it is not possible to know in advance which products, and in what quantities, will be produced. Thus, using demand to plan production is technically ineffective and by implication, economically inefficient. This notion is now examined.

The Production Process in the Timber Industry

One characteristic of most manufacturing operations is that there is a fixed relationship between the raw materials and the finished product; for example, a chair has one seat, one back and four legs and we can calculate quite precisely the raw material requirements to assemble one chair – planning the production of many chairs is a simple mathematical exercise. Here, multiple components produce one finished product, a process called aggregation, and this is the theoretical basis of Materials Requirements Planning (MRP) and the more comprehensive Manufacturing Resource Planning (MRPII).

Another class of manufacturing, called disaggregation, manifests itself where one item is broken-up into multiple end products. This can be further complicated in that the outputs may not be able to be perfectly pre-determined. An example of this class of manufacturing is timber milling; due to inherent and invisible imperfections in a log, it is not possible to know exactly what finished products will be delivered until the milling process is completed. These are fundamental differences and MRPII does not cater to the requirements of this class of manufacturing (Porter, Little et al., 1999).

The extended timber supply chain encompasses silviculture and harvesting operations in the plantation, milling, treatment and finishing operations at the sawmill, and warehousing, sales and delivery operations in the distribution business. A cycle of this supply chain can take 25 to 40 years with silviculture decisions, for example, having a direct impact on the characteristics of the sawlog delivered as raw material to the sawmill 25 years later. The key area of focus in this paper, however, is the milling process and how the production capability is used operationally to meet actual demand. There are three major stages in the production of timber: (1) milling in the green mill; (2) drying in the kilns; and (3) planing, treating and packaging in the planer mill, as shown in Figure 1 – Milling Process Schematic.

In the green mill, sawlogs of similar characteristics, most notably diameter and length, but increasingly projected finished good quality determined using advanced techniques such as sonic testing, are sorted together as a batch called a log sort, debarked and scanned to determine a precise measurement of each log. A cutting pattern, specifying the way the log should be cut to achieve a desired finished good outcome, is used to guide the milling of the sort type. At best, about a 70% accuracy in outcome can be achieved over a population of sawlogs in the sort type due to imperfections that become apparent as the log is progressively milled, and the skill of the saw operator is a significant element in the process. At the end of the green mill process, products of similar dimension are sorted into bins. Products may be sold at this stage as un-moulded green timber, or go directly to the planer mill to be moulded, but most is stacked in fillets to be kiln dried.

Figure 1 – Milling Process Schematic

Similar green fillets are aggregated into a kiln charge. Factors such as the dimension, moisture content and degree of straightening required of each batch is used to determine the placement of each charge within the kiln, and the temperature and time of the drying process. Weights are placed on top of the stacked charges to assist in straightening the timber as it softens at high temperature. Treatment time can range from 12 to 48 hours in the kiln. Depending on weather conditions, the timber may be used immediately or it could be further seasoned for up to three weeks.

In the planer mill, each fillet is broken apart and fed through the mill to be planed to specific and exact dimension in a process called moulding. Major imperfections such as knot-holes or intruding bark cut are cut out, and the shorter pieces may be rejoined with finger-joints and adhesive. Each piece is stress graded by machine to determine its Machine Graded Pine (MGP) rating, a measure of strength, and the rating stamped on the piece. Pieces with appearance defects are also sorted out manually. Some wood is treated with sprayed-on chemicals to inhibit fungal or insect attack depending on the intended application. Finally, pieces of similar quality and dimension are sorted together and packaged, labeled and plastic wrapped.

Although an apparently simple process, operational constraints on the delivery of sawlogs, varying quality of those sawlogs, the range of product, choice of dimension, size of order, number of orders, seasonality of demand and delivery alternatives, all make for a complex planning problem, and computer support is required.

Production Planning Support

Enterprise Resource Planning (ERP) computer systems are implemented in a company to provide an enterprise-wide, integrated and real-time view of the operations of the company with respect to financial and operational metrics. With functionality than can cover most of the operations of a company, including order-taking, production planning, delivery planning, and billing, it is not surprising that ERPs are used to facilitate the (re)-implementation of a company’s supply chain. A fundamental issue, however, is that ERP software has typically been developed for discrete manufacturing (Harrold, 2000) using the concepts of MRPII.

While the kiln and planer mill processes are predominantly aggregation, the green mill process is disaggregation, and further, there is a large degree of unpredictability in the output from the process. This presents a significant difficulty to planning across the supply chain, with some ERP implementations starting at the kiln and moving forward, not at all cognizant of the green mill or of the forestry processes preceding that. The major input cost (the sawlog), and anecdotally, the two most significant opportunities to improve the value of the sawlog (the cut-to-length in the forest and the first cut in the green mill) are simply outside the visibility of supply chain planning.

While much research has been undertaken examining discrete manufacturing, process manufacturing has been less explored (Jahangirian and Conroy, 2000). Current research efforts appear to be focused on genetic algorithms as a way of predicting outputs in “dynamic-stochastic flow shops” (Jahangirian and Conroy, 2000), however at this time the results are inconclusive. It is only recently that the issues apparent in planning production in highly dynamic, relatively complex, and probabilistic environments (for example, a timber mill) have attracted academic attention. Jahangirian and Conroy (2000) report on a study of 170 research articles in the scheduling domain between 1952 and 1994. In that study, there was no evidence of papers in the “dynamic-stochastic category of flow-shops” (for example, a timber mill). Jahangirian and Conroy (2000) went on to apply learning machine (genetic algorithm) scheduling strategies to the problem. While their test results on a basic model showed how the system adapted to dynamic patterns and converged to an optimum solution, Choudhry (2000), for example, was able to report only mixed results in a research study integrating constraints management and genetic algorithms using extensive production data. Porter and Little et al (1999) note, however, that “the trade-off decisions involving resource availability and/or utilisation, against operating costs, against customer service, and probably against logistics costs, are so complex to compute that an exact solution is worthless anyway”. By the time a solution can be computed, the problem has long passed.

Nevertheless, process manufacturing companies continue to make product and to seek ways to perform better. In the absence of optimising computer support for production planning, the mill production planner must continue to use the heuristics developed over a 100 years ago (Williston, 1981). It must be concluded, therefore, that each process manufacturer is left to analyse their own needs, and to select software that best meets these needs. This level of analysis, however, is probably beyond the resources of many companies and a framework to support the selection and implementation of a production planning and scheduling system is required (McGarrie, 1998). In the absence of such a framework, the manufacturing department is probably obliged to adopt the ERP selected for the broader business and asked to adapt its processes and philosophies to the needs of the software, software which fundamentally may not be able to support the manufacturing operations of the company. At best, experimental work to find a deterministic production planning and control solution with application in the timber milling industry is continuing, and it is unlikely any commercially usable techniques will become available in the near term.

A review of the literature seems to support the notion that the timber industry is production-orientated because of the technical difficulty of matching production to demand. Standard ERP toolsets as used in many integrated enterprises to facilitate the implementation of supply chain planning do not cater to the demands of a dynamic stochastic flow shop, and current research would not appear to offer any near term solutions.