Printer Friendly

Gang Rip Saw Arbor design.

Researchers at North Carolina State University have developed a software program for helping optimize yields of rip-first rough mills employing gang ripsaws with fixed saw positioning.

A manufacturer considering a rough mill system must decide whether to employ cross-cut or rip-first processing, the level of automation and the degree of computer assistance involved. Computers can make real-time decisions that affect the processing of each board as manifested in any of the "optimizing" systems for cross-cut and gang rip rough mills. In contrast, computerized decision support systems are used to make complex decisions that are required less frequently. This discussion overviews a decision support system for designing gang rip arbors and its application in gang ripping operations. It shows that by combining real-time and off-line computer assistance, gang ripping using non-removable saws can be made more yield efficient than previously possible.

GRADS (Gang Rip Arbor Design Software)

GRADS is a decision support system for IBM and compatible computers that can improve the yields obtainable in rough mills using gang ripsaws with fixed position saws. An important decision in implementing a rip-first rough mill is whether to employ a gang ripsaw with fixed or moveable saws. The choice can be influenced by the value of the lumber, the production volume required, the degree to which rip widths can be standardized, the mix of random versus specified width rippings, the required quality of glue joints, production lot sizes and the capital available to install the system.

Both types of systems are currently used in various configurations and the relative merits of each are debated. In general, gang rip systems featuring moveable saws are more expensive, more automated and may not produce glue joints of comparable quality to fixed position saws. The primary advantage of moveable saw systems is maximum yield attainment through dynamic adjustment of the saw spacings to fit each board and its defect pattern. Besides adjusting the ripping pattern to use the entire width of the board, an operator can also embed collinear defects in one strip just wide enough to remove the defects. This longitudinal defecting ability is unique to moveable saw systems.

Gang rips with fixed saws offer lower initial cost, high-quality glue joints and yields that can approach those obtained with moveable saws. The issue here is not which system is better; both have appropriate applications. Rather, the use of GRADS to design arbors for "optimizing" fixed saw gang rips increases the yield obtainable and alters the evaluation process for those considering a rip first strategy.

Yield improvement through GRADS

Most gang rip systems with fixed position saws modify the position of incoming boards to achieve efficient ripping patterns. Combining the power of GRADS for designing an "optimal" set of fixed position saw arbors with intelligent board positioning yields previously unattainable become possible. Although the design of an efficient arbor is a critical yield determinant with fixed saw spacings, the analytical complexities involved in generating good arbors has limited the development of software to address this problem. Consequently, arbor design is often based on operator judgement or experience.

An arbitrarily designed arbor can produce good yields, but potentially better arbors that produce higher yields will usually exist. However, it is a difficult process to identify these improved arbor configurations. The combinatoric nature of the problem disallows manual solution, creating the need for an analytical tool that consistently delivers "optimal" saw arbor designs. To illustrate, for the case of five rip widths arranged on a 24-inch arbor, there are over a million possible combinations of saw spacings. The problem grows even larger when relative priorities for each rip width are considered. The GRADS system intelligently evaluates potential combinations of saw spacings and rip width priorities to produce a "recommended" set of arbors. These arbor configurations produce high yield, the desired quantity of each rip width and a low number of arbor changeouts.

This software augments the yield potential of gang rip "optimizing" systems that provide real-time positional adjustment to align each board with different saw spacings. This "optimizing" technique is often implemented with laser-line lighting adjusted by an operator to indicate maximum usable board width. A computer then considers cutting bill quantity requirements and any special rip width priorities in determining the recommended rip pattern. It directs an operator or controls a moveable fence to position the board so that the optimal ripping pattern is produced. This combination of optimal arbor design through the GRADS software and the real-time board positioning can result in a high yield, high-volume system.

Data requirements and assumptions

The GRADS system requires the user to supply a cutting bill which consists of up to eight rip widths and the board footage required in each width. The user must also specify a distribution of board widths to insure that arbors are generated based on realistic data tailored for any lumber grade (or grade mix) being processed. A maximum arbor width is specified, normally between 12 and 24 inches. A saw blade thickness of 1/8, 3/16 or 1/4 inch must also be provided to account for kerf. The GRADS software assumes that all rip widths are edged on both sides by saws, i.e., the number of saw blades exceeds the number of rip widths by one.

Software objective

Using the data provided, plus internally generated weights for each rip width, GRADS seeks to generate an efficient set of arbors that maximizes yield and keeps the number of arbor changeouts low. Yield is defined to include the sum of the footage processed. The system estimates the amount of lumber needed to fit the cutting bill and guarantees that all rip width demands will be satisfied. It also recommends arbors that hold overrun to a minimum by dynamically adjusting the weights associated with each rip width. In this way, GRADS seeks to simultaneously satisfy the practical requirements of high yield, infrequent arbor changeouts and low overproduction of ripping widths.

A sample session with GRADS

GRADS is a menu-driven system that employs an "improvement" algorithm that generates an initial solution and presents it to the user. The operator can accept the solution or ask the computer to spend more time looking for an improved solution (higher yield). At each iteration, the computer searches many possible arbor configurations and compares them to the current solution. As soon as an improved solution is found, that arbor configuration is displayed and the operator can either accept it or continue to look for a better arbor set. A sample session is now described based on a system using a 24-inch saw arbor and 1/8-inch saw blades.

The GRADS system first presents the user with a data input screen. The user enters the demand (board footage) he requires beside each needed rip width, presented in 1/8-inch increments. Rip widths from 1 inch to 4 7/8 inches are shown. By hitting the F8-Page Ahead key, a second screen appears that accepts demand for rip widths between 5 and 8 inches.

After the cutting bill is entered, hitting the F6-Board Stock key brings up an input screen accepting data to describe distribution of board widths in the lumber to be processed. Acceptable board widths range from 3 to 11 inches in 1/16-inch increments. Hitting the F8-Page Ahead key brings up a similar screen for boards ranging in width from 7 to 11 inches. The user enters quantities specified in lineal feet that can be based on a manual sampling of incoming lumber, or on data collected automatically by many of the optimizing gang rip systems commercially available. Once the data characterizing the cutting requirements and the width distribution of the lumber has been entered, the user hits the F10-Run key and GRADS begins the process of generating potential saw arbors for consideration.

Referring to the Figure 1, the user sees the Current Status screen that displays the current solution in terms of the recommended set of arbors and performance statistics. In this example there are five rip widths shown under the heading "Finished Widths" with the number 1 mapped to a 1-inch width, 2 to a 1 1/2-inch width, 3 to a 2 1/4-inch width and so on. Under the heading "Currently Recommended Arbors," the arbor configurations are indicated by the arrangement of the numbers associated with each rip width. The first arbor consists of 16 1-inch rippings, the second consists of 14 1 1/2-inch rippings, the third arbor uses 10 2 1/2-inch rippings and so on. Note that in this initial solution, each arbor consists entirely of rippings of a single width. Not surprisingly, there are five arbors recommended, equalling the number of rip widths in the cutting.

The strange looking set of five numbers under each arbor are the weights assigned to each of the five rip widths being processed. Initial weights are assigned in relative proportion to the width of each ripping. For example, the 1-inch width has a weight of 0.833, the 1 1/2-inch width has a weight of 1.250 (0.833 x 1.5), and the weight of the 3 1/2-inch width is 2.917 (0.833 x 3.5). As the first arbor in a set begins to satisfy demand in each width at different rates, these weights are dynamically adjusted to insure that subsequent arbors in that set will not cause excessive overrun. To the right of each recommended arbor, the yield for that arbor is shown under the heading "Usable" and the expected lineal footage of lumber to be processed while that arbor is active under Run Time. The Run Time statistic allows the operator to disregard an arbor as if its required usage is very low. It would not be economical to swap out an arbor if its run time and effect on overall yield were negligible.

The Figure 1 solution represents a starting point in balancing the issue of yield improvement and minimizing overrun. It is worth repeating that the Portion Usable (yield) calculated by this program includes any overrun generated. By hitting the F1-Profile key, a Profile Window will pop up (outlined in bold black) allowing the user to monitor the level of overrun generated in each width. Since the current arbor set includes five arbors, each one producing only a single width, no overrun (Requested amount equals Expected amount) in any width is observed.

Referring again to Figure 1, the user sees the expected amount of lumber needed to satisfy the cutting bill (Board Feet Total), the amount of waste generated (Board Feet Waste), and the expected yield (Portion Usable) in the lower left box. Note the current yield of 73.2% was achieved using five arbors and generating no overrun. Subsequent recommended arbor sets will improve yield, often reduce the number of arbors used and potentially generate a small amount of overrun in some width categories. GRADS' primary objective is to improve yield, though the software is also effective at holding the required number of arbors low and minimizing the amount of overrun generated.

By continuing to hit the F2-Generate key, the user can tell the computer to continue generating better solutions. A computer with an 80386 or 80486 processor will produce the next solution almost instantaneously for the first several iterations. Figure 2 shows the solution obtained after only four iterations. Note that the yield has improved to almost 88.8% and the number of arbors needed has been reduced from five to three, a very desirable result. By hitting the F1-Profile key, the user can see as in Figure 3, that overrun is limited to the 1-inch and 2 7/8-inch widths. The demand for the other three rip widths is expected to be exactly satisfied by using each arbor for the indicated Run Time.

The solution procedure converges very quickly to a yield that is difficult to improve on. In this example, yield was 88.8% on the fourth iteration and by the 34th iteration the yield had only increased to 90.3%, a figure that is almost a limit due to the waste caused by a kerf. To illustrate, if an 8-inch-wide board is ripped into two 1-inch strips, one 2/14-inch strip and one 2 7/8-inch strip using 3/16-inch saws, the kerf waste is 11.7%. If the user asks the computer for a better solution, it would proceed to evaluate new arbor configurations and compare them to the yield of the current solution. If no better solution is found in the next 10,000 potential arbors it evaluates, the user is notified of this fact and is prompted to either terminate or continue the search for a better arbor. Once the solution process reaches this stage, a 33 MHz 80486 computer requires about 40 seconds to evaluate 10,000 additional arbors.

A user can manually enter an arbor and a set of weights for evaluation by the algorithm. By hitting the F9-Try Arbor key, a pop-up window will appear, allowing the user to enter an arbor configuration and the associated weights for each rip width. By entering a "good" arbor, the user can specify a starting point for the algorithm that is more efficient than the initial solution shown in Figure 1.

Conclusions

GRADS is a decision support system that can improve yields in rough mills that employ a rip-first strategy implemented with fixed position saws and computer optimization for board positioning. For the example included in this article, the ripping yield obtainable from arbors generated by GRADS exceeded 90% suggesting that only marginal yield improvement could be gained by substituting moveable saws for fixed position saws. Used in this way, GRADS can allow a manufacturer to evaluate the effectiveness of non-moveable saw gang ripping in his operation.

As a design tool, GRADS can generate very efficient saw arbors and significantly improve yields over arbitrarily designed arbors, particularly for wide arbors of 24 inches. The combination of GRADS for arbor design and "optimizing" software for board positioning significantly increases the yield potential of gang rips with fixed position saws. This creates an inducement to consider, and a tool to help evaluate, the use of fixed position saws for gang rip rough mills.

For more information concerning GRADS or the Furniture Manufacturing and Management Center contact: Tom Culbreth, director, or Lee Houston, furniture extension specialist. Write to: The Furniture Manufacturing and Management Center, North Carolina State University, Box 7906, Raleigh, North Carolina 27695 or call (919) 515-3335.
COPYRIGHT 1993 Vance Publishing Corp.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:computer program for gang ripping operations
Author:Fathi, Yahya
Publication:Wood & Wood Products
Date:Feb 1, 1993
Words:2407
Previous Article:Postformed edge not just for countertops anymore.
Next Article:Taking the mystery out of furniture manufacturing.
Topics:


Related Articles
Diamonds in the rough mill.
Highlights from Ligna.
Developments in panel optimization.
Computer software uses keep growing.
Panel saw optimization: minimizes waste, maximizes yield.
Woodcraft optimizes rough mill productivity.
Rough mill operations watch their waste.
Modernizing the rough mill.
IWF 2000 Breaks Attendance Records.
Ligna 2001 Draws a Global Crowd.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters