Variation is the Theme
Reducing moisture content variation in your shipments of KD lumber needn’t be expensive. In fact, given new and emerging lumber markets, it pays!
by François Léger, Forintek Canada
Moisture content (MC) variation in lumber delivered to clients represents a major problem for the lumber industry. This is a quality parameter that is hard to control, but it adds to production costs and causes client dissatisfaction. In the current competitive climate, reducing MC variation is no longer a luxury; it has become a necessity.
This situation is due to changes in the marketplace. Not so long ago, the production of our sawmills mostly served the needs of residential construction. Today, residential construction only uses some 35% of all shipments (see Fig 1.). Needs have changed and, inevitably, industry must change accordingly.
The renovation market is one example. The expansion of this market has been such that it has led to the adoption of new in-house lumber grades in the mills. We now have “Premium” and “Select” grades, among others. Is this recognition of appearance going to be enough? No, it won’t. To date, MC variation has mostly been neglected as a factor in lumber quality.
The need for MC control is real. Grading rules were set some years ago, mostly in relation to the needs of the building construction market. By today’s standards they definitely seem too lax and too broad. Worse, they have contributed to wood being perceived as dimensionally unstable. In other words, lumber develops twist and other forms of warp at the client’s shop, and the whole industry suffers as a result.
By limiting MC to 19% or less at the mill without limiting variation, current grading rules have become a weak point in the system. Research shows that twist can double when lumber MC drops from 20% to 14%. Lumber continues to dry between the mill and the client, and this may explain some disagreements with clients regarding standard conformance. In new markets, as well as residential construction, good MC management will become a differentiating factor in determining client satisfaction.
Lower costs too
Reducing final MC variation does not necessarily result in inflated production costs. Actually, at Forintek we believe it’s the opposite. Reducing losses due to excessive MC variation is not only a matter of market sense, it makes economic sense as well, once you factor in the true costs of poor MC control. First, you need to pull out all underdried lumber. Mills usually do a pretty good job of this, but some kiln wets still find their way to the marketplace. Over-dried lumber is more likely to have warped, and this translates into financial losses due to degrade. Losses due to degrade grow exponentially as final MC goes down. We must also factor in losses resulting from reduced productivity at the planer mill when over-dried lumber is processed. When we combine the losses due to degrade, lower productivity, and kiln wets, we end up with a graph that describes the financial impact of MC variation in mills (see Fig 2).
Clearly, everybody tries to dry lumber to within MC levels considered good practice, such as between 11% and 19% under current conditions. But with what results? MC levels measured in industry after drying vary greatly. The results indicate that 25% of the pieces still fall outside the range considered good practice (More info on current kiln practices can be found in An Evaluation of Wood Kiln Control Practices by François Léger and Mouloud Amazouz, and available at www.forintek.ca and http://cetcvarennes.nrcan.gc.ca. Based on a survey involving over 100 kiln operators, quality controllers, superintendents, etc.)
If we apply the typical average MC distribution illustrated in Figure 3 to a 100 MMbf mill and total up losses due to degrade, wets, and productivity in a curve of the type shown in Figure 2, we can demonstrate that total losses are in the order of $750,000 per year. These losses are reflected in performance indicators such as reduced recovery of No. 2 & Better and greater trim losses.
In our evaluation, we include losses associated with under-dried lumber even though standards are rather permissive in this regard. Our rationale is that lumber shipped at a MC level exceeding equilibrium moisture content (EMC) at the client’s place will experience additional warp, which will cause buyer dissatisfaction. As a way of quantifying such dissatisfaction and the resulting difficulty in selling lumber containing kiln wets, we treat them as “degrade” even though the pieces would not necessarily fall into this category due to the 5% tolerance allowed by NLGA.
To avoid these losses, you need to monitor and control the process, and this was the objective of the pilot project undertaken by Forintek with Scierie Leduc, a member mill. The project has been named “Sirocco.” It aims to monitor production of individual bundles to identify best drying practices. This involves correcting deviations that cause financial losses for the company, and thus requires detecting and diagnosing process errors in the first place.
From theory to practice
What follows is our experience with implementing this innovative method and applying principles to real-time application. It took a great deal of effort and tenacity from the mill’s production team, including Jean-Pierre Gagné, general superintendent, Normand Boucher, planer mill foreman, Pierre Lemieux, kiln operator, and Keven Perron, instrumentation and automation guru.
There were two main challenges. First, we had to quantify the losses due to MC variability as illustrated in Figure 2. All mills are different in this respect, as their performance varies with the process and the wood supply. The second challenge was to determine, in real time for each bundle, a histogram of its MC distribution, and to identify what conditions led to bundle MC distributions falling either within or outside the prescribed limits. The idea was to use this information to capitalize on good results and thus improve drying practices.
In this particular situation, our objective is to provide a visual tool to monitor the number of under- and overdried pieces. Figure 4 shows a diagram of a load of lumber colour-coded based on final MC. If, for example, a bundle contains more over-dried pieces, it appears red. On the other hand, if it contains more under-dried pieces, it appears green. The intensity of the colour relates to the proportion of “off-spec” material. If it happens to contain a mix of under- and over-dried pieces, this will translate into a mix of red and green, turning to black in the worst scenario. If it is within the pre-determined limits, the bundle appears white. This is the ideal condition and, more importantly, the best condition to reduce production costs.
From Figure 4, we are in a position to relate the results obtained at the planer mill to the history of the load before, during, and after drying. Specifically, the Sirocco project allows us to associate a bundle to its position in the kiln, the details of the drying process, yard storage times before and after drying, and all other process characteristics right back to the sawmill. This is made possible by bundle traceability, the key ingredient in the success of this method.
In this article we have considered MC variation control exclusively from the point of view of production costs, which is the most effective way to achieve acceptance of new drying practices at this time. However, it should be kept in mind that to achieve client satisfaction will require further efforts on reducing final MC variation, and tools such as Sirocco can facilitate this process.
The Sirocco project is the joint result of the expertise and skills of many individuals from Forintek Canada Corp., CANMET-Varennes (CANMET Energy Technology Centre, Natural Resources Canada), Forac (a research consortium on value creation networks), OSIsoft Canada (provider of the RtPM, Realtime Process Management platform) and, obviously, Scierie Leduc (a softwood lumber mill).
François Léger is a drying researcher with Forintek Canada Corp. based in St. Foy, QC. You are welcome to discuss or comment on this project by contacting the author at francois.leger@qc.forintek.ca.
