### WSO™ Sawmill Optimization Example

The following is presented as an example of how WSO™ can be used to optimize a sawmill, and also of the type of reports that can be produced from WSO™.

The model is as follows:

- A single two-line small-log sawmill, and associated woodlands operations and log sales, are simulated. Multiple mills could have been included.
- Log volumes are recorded in m3, since a hypothetical Canadian mill is being modeled, but it could equally easily have been set up to record log volumes in MBF of Scribner scale, or in green tons.
- The timber supply consists of three Spruce-Pine timber types (Timber Models), with names tmSml, tmMed and tmLge, aggregated from data for 70 strata. Some Douglas Fir and Balsam Fir are present.
- There are three separate timber supplies, tsA, tsB and tsC, characterized by increasing average hauling cost to the mill, for a total of nine combinations of timber type and hauling cost. tsA represents timber that is closest to the mill, tsC represents timber that is farthest from the mill.
- Logging is assumed to be modified tree-length, with the production of pulplog, small-, large- and oversize-sawlog sorts and a peeler sort. The peelers and oversize logs and all Douglas Fir sorts are assumed to be sold. There are separate sorts for Balsam Fir.
- For the purposes of analysis, the small-sawlog sorts are subdivided into “virtual sorts”, with various minimum butt diameters, which may or may not correspond to actual log sorts. The purpose of the “virtual sorts” is to determine where the division between pulp and sawlog sorts should be, and to determine optimum allocation between mills when modeling of a second sawmill is implemented. There is provision in the model for easy implementation of cut-to-length harvesting.
- The sawmill has a small-log line, with an 8” four-sided canter and close-coupled VSA gang, and a large-log line, with a canter-quad and curve-saw gang. There is one edger and one trimmer for the two lines. All centre-cant boards pass through the trimmer, after which they may be re-edged and pass through the trimmer again.

Sawmill Flow

- The model has been set up to run the merchandising and sawmill for two 8-hour shifts, i.e. 80 hours per week of regular time, with no overtime or weekend time. The kilns and planer are assumed to run for only as long as is required to finish the lumber. No 1” lumber is produced and there are no sales constraints.
- Under these circumstances, the marginal breakeven values calculated by WSO™ for the various rough lumber items are the weighted values of the lumber grades produced, less the variable costs of drying and planing. Balsam Fir is assumed to be run through the sawmill separately from the Spruce and Pine which are run together.

The operation of the sawmill optimizers is simulated by using SAWSIM^{®} to select the sawing pattern to be chosen on the large-log line, and to calculate the corresponding lumber yields.

- In Run 1_1 the canter optimizers are assumed to be set up to maximize the lumber value per log. Under this circumstance, the sawmill production is limited by the trimmer. The canter-quad runs to capacity, and there is spare capacity on the small canter line. This means that there are non-zero marginal breakeven values for canter-quad and trimmer time, and the marginal breakeven values of zero for time on the other machines.
- Run 1_2 is similar to Run 1_1, except that costs of canter-quad and trimmer time are increased by the marginal break-even costs (opportunity costs) for time on these machines when calculating the values of the sawing solutions. This causes sawing methods to be chosen which produce fewer boards through trimming and re-edging. The result is that load is taken off the trimmer, so that the small-log line is able to run to capacity. A small amount of spare capacity develops at the trimmer. The overall lumber recovery drops, but the lumber production and the objective function, i.e. the margin between income and fixed and variable expense, increase. Thus Run 1_2 maximizes the profit per hour instead of the value recovery per log.
- In this example the objective function, i.e. the net income before taxes, was increased by $1.2 million per year.

Had there been constraints on lumber or chip sales, or in the lumber finishing, these would have affected the marginal breakeven values for lumber and sawmill processing time. It would then have been necessary to use these modified values when calculating the values for each sawing pattern and determining the patterns to be chosen by the canter optimizers.

Firm conclusions should not be drawn from this example. For instance, had different lumber prices been used in the example, the results could have been substantially different.

Please click here to view Excel results. Side-by-side reports for the two runs are displayed.

Please note that the Excel reports were generated by HalcoMiner™ from an MS Access database file generated by WSO™.

- To view a report click on its title in the Contents sheet and then click on "Return to Contents" to view another report.
- Please contact HALCO if you have questions, or would like more information or other examples.