AN ANALYSIS. oFTFHF' OPERATING FFOLICFIAR’Djf} V f , :2: - "PROCEDURES OF A FIRM PRODUCING DOMESTlcgj'V;;?;;:;;J;;§:5342:??? WALNUT VENEER ’ ‘ ‘ “ " Thésis for thé Dégreie D! Ph D “CW" 37"“ ”WEAR" . ; MONTE RALPH HAROLD z; , - 1972 “ ABSTRACT AN ANALYSIS OF THE OPERATING POLICY AND PROCEDURES OF A FIRM PRODUCING DOMESTIC WALNUT VENEER by Monte Ralph Harold The nature of any firm which converts timber into solid wood commodity—type products is unique because quality and quantity yield of the raw material is not easily determined in advance, since quality and quantity—reducing characteris— tics are often hidden. Quantity is estimated by the use of log rules which may or may not prove accurate. Quality is generally estimated by the use of 10g grades which have proven satisfactory for production of some wood products — hardwood factory lumber is one successful application. Great demand for walnut wood products accompanied by a concurrent shortage of walnut timber has set the price structure for this species at a very high level. An average price per thousand board feet has been reported as high as $1,367 for prime log 24 inches in diameter and larger. In contrast, the price for yellow—pOplar, a non—decorative Monte Ralph Harold species, is reported at $171 for equivalent quality level and unit volume of this material. These relatively high unit prices and the lack of adequate methods for quality determination present a most difficult raw material acquisi- tion problem for domestic veneer producers. Raw material acquisition problems facing domestic walnut veneer producers became apparent through personal contact with a specific firm. It became obvious there was a general lack of confidence by management and 10g buyers in their current log grading system. A decision was made to study this problem in depth by gathering and analyzing data and Operational information from this walnut veneer—producing firm. The objective was to critically analyze the purchasing, processing and marketing system within this firm. The pur- pose was to search for methods by which improved efficiency of the system would result. A second objective was to test the use of modern analytical procedures and statistical methods in an effort to predict the unit value of resulting veneer from the exterior characteristics of the log. The approach was to carefully select the study material over a range of physical sizes, quality levels and geograph— ical ranges; collect the necessary raw material, conversion Monte Ralph Harold process and final product data; convert the data to useable form; develop a statistically sound predictive model; analyze the results and state the conclusions. Two multivariate statistical methods, multiple regres- sion and discriminant function analysis, were used in an effort to develop a predictive mathematical model. The ability of the regression model to account for only 13 per- cent of the variability of unit veneer price is considered to be of little consequence. Model improvement techniques also failed to significantly improve the model. Inability to develop a significant equation for distinquishing between grade classification by discriminant function was interpreted as failure of this method to solve the problem. Conclusions of why these efforts failed fall into two general classes: (1) marketing system limitations, (2) buying system limitations. The absence of objective grading rules for walnut veneer may be a major marketing system limitation. If pro— duct quality can not be stated in quantified terms, quality yield may vary over time and between output measurement units without being recognized. This standardization is a prerequisite to developing and maintaining a predictive out— put value model based on standardized input information. Monte Ralph Harold It was concluded that inexperience may have resulted in several marketing weaknesses. Evidence was presented that veneer may be underpriced, especially the higher quali— ty levels. The veneer price range for the firm was 1.0 to 8.0 cents per square foot; while the competitor firm price range was from 3.0 to 60.0 cents. The firm has identified few specialized veneer markets; lacks the experience to identify these veneers; and fails to hold uniquely figured flitches in inventory to meet market opportunities. The export of walnut veneer to Europe is one of the markets to which the firm could respond, for high prices paid for export logs has been a factor in the competi— tive nature of log buying. Certain buying system improvements warrant consideration. They are: (l) more thorough training of log buyers, (2) developing an incentive plan based on profits from logs purchased by the individual log buyers. Monte Ralph Harold It was concluded that inexperience may have resulted in several marketing weaknesses. Evidence was presented that veneer may be underpriced, especially the higher quali- ty levels. The veneer price range for the firm was 1.0 to 8.0 cents per square foot; while the competitor firm price range was from 3.0 to 60.0 cents. The firm has identified few specialized veneer markets; lacks the experience to identify these veneers; and fails to hold uniquely figured flitches in inventory to meet market Opportunities. The export of walnut veneer to Europe is one of the markets to which the firm could respond, for high prices paid for export logs has been a factor in the competi— tive nature of log buying. Certain buying system improvements warrant consideration. They are: (l) more thorough training of log buyers, (2) developing an incentive plan based on profits from logs purchased by the individual log buyers. AN ANALYSIS OF THE OPERATING POLICY AND PROCEDURES OF A FIRM PRODUCING DOMESTIC WALNUT VENEER by Monte Ralph Harold A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Forestry 1972 ACKNOWLEDGEMENTS The author expresses the greatest of appreciation to Dr. Otto Suchsland, Chairman Of the Guidance Committee, for his guidance and inspiration. Acknowledgement of his talent to inspire one to persist over adversity has had special meaning to the author through this study and his entire graduate program. I am grateful and indebted to Dr. Aubrey Wylie who helped make available the Opportunity to do the field work for this study and for his guidance in the early stages of the study. The author also expresses a realization of indebtedness to Dr. Richard Gonzalez, Dr. Alan Sliker and Dr. Eldon Behr the other Guidance Committee members. Special appreciation goes to Mr. Gene Freeman who COOperated so fully during the data-gathering phase of the study. I am also grateful to the McIntire-Stennis COOpera— tive Forestry Research Program for its financial contribution. I wish to thank my wife Carol and to my children Eric, Kristen and Scott who seldom complained about my many hours of study. ii LIST OF TABLES TABLE LIST OF FIGURES . . . . LIST OF APPENDICES . . Chapter I INTRODUCTION II III IV THE TIMBER QUALITY CLASSIFICATION OF CONTENTS PROBLEM . . . . . . . . A. General Problem . . . . . B. Veneer Log Evaluation . . THE DOMESTIC VENEER LOG GRADING PROBLEM . . . A. B. Study Firm 1. General Quality Levels 2. Domestic Log Grades . Objective 0 THE STUDY FIRM'S ACQUISITION, PROCESSING AND MARKETING SYSTEM A. B. C. Acquisition Processing Marketing 0 O O O O O PROCEDURE, SAMPLING AND DATA COLLECTION . A. B. General . Procedure 0 O O O o 0 iii Page vi vii p 26 26 27 3O 39 41 41 44 47 49 49 49 Chapter Page C. Sampling . . . . . . . . . . . . . . . 51 D. Data Collection . . . . . . . . . . . . 55 1. Measured Variables . . . . . . . . 56 a. Continuous Measured Variables . 56 b. Discontinuous Measured Variables . . . . . . . . . . . 57 2. Ranked Variables . . . . . . . . . 57 3. Attributes . . . . . . . . . . . . 6O 4. Computed Variables . . . . . . . . 60 E. Defect Recording System . . . . . . . . 61 F. Yield Data . . . . . . . . . . . . . . 65 VI PRELIMINARY COMPUTATION OF FIELD DATA . . . . . . . . . . . . . . . . . . . 68 A. General . . . . . . . . . . . . . . . . 68 B. Log Characteristics . . . . . . . . . . 68 C. Log Volume Characteristics . . . . . . 69 D. Clear Area Cuttings and Clear Volume Computation . . . . . . . . . . . . . . 71 VII STATISTICAL EVALUATION . . . . . . . . . . 78 A. General Estimator . . . . . . . . . . . 78 B. Regression Methods . . . . . . . . . . 80 1. Confidence Limits . . . . . . . . . 84 a. Standard Error of Estimate . . 84 b. Coefficient of Multiple Determination . . . . . . . . . 84 c. Coefficient Significance Tests . . . . . . . . . . . . . 84 2. DevelOping the Best Equation . . . 85 a. All Possible Equations . . . . 86 b. Backward Elimination . . . . . 87 c. Forward Selection . . . . . . . 87 iv Chapter VIII Dummy Variables Model Testing . Results . . . . mun-bu) Variable Transformation . . C. Discriminant Function . . . . . 1. Model Testing . 2. Results . . . . ANALYSIS OF RESULTS AND A. Background . . . . B. Analysis of Results C. Conclusions . . . . 1. Marketing System Limitations CONCLUSIONS a. Veneer Grading . . . . b. Pricing . . c. Specialized Markets . . d. Supply—Demand Disruptions e. Innovative Action . . . 2. Buying System Limitations . a. Grading Half—lOgS . . . b. Buyer Training . . . . LITERATURE CITED . . . . . . . . APPENDIX Page 88 88 9O 94 101 105 106 110 110 110 112 112 112 113 118 119 120 120 120 121 122 125 LI ST OF TABLES Table Page 1. Summary of veneer characteristics and defects of premium grade and good grade plain sliced walnut veneer . . . . . . . . . . 10 2. Monthly profit variance report for domestic veneer log grades . . . . . . . . . . 38 3. Summary of model testing results . . . . . . . 96 4. Regression coefficients Of independent variables significant for predicting price per square foot of walnut veneer . . . . 100 5. Results of the discriminant analysis test of two grade, eight variable model (model II—A) . . . . . . . . . . . . . . . . . 107 6. Results Of the two grade, two variable discriminant model (model II—B). . . . . . . . 108 7. A comparison of the differential pricing policies of the study firm and a competitor . . . . . . . . . . . . . . . . . . 116 vi Figure 10. 11. 12. LIST OF FIGURES Forest Service standard specification for hardwood factory lumber logs . . . . . . . Wood quality and the log cost assessment problem . . . . . . . . . . . . . . . . Adventitious bud cluster and knot as defect indicators on a log surface . . . Adventitious bud cluster and knot defects in walnut veneer . . . . . . . . . . . Bird peck holes in log bark surface . . . Bird peck as color distortion on the end Of a log . . . . . . . . . . . . . . . . . Veneer yield value as a function of log diameter for a stratified log quality sample . . . . . . . . . . . . . . . . . . Premium quality logs such as this are exported to EurOpe for furniture veneer production 0 O O O O O O O O O O C C O O . Medium quality logs as illustrated here are processed domestically into flatésliced veneer . O O O O O O O O O O O O O O O O 0 Lower quality walnut logs such as these are sawn into lumber . . . . . . . . . . . . . Domestic walnut veneer log grade rules . . Flowchart of study firm's Operating system . . . . . . . . . . . . . . . . . . vii Page 12 15 17 19 22 23 29 32 34 36 43 Figure Page 13. Schematic drawing of log to flitch conversion process . . . . . . . . . 46 14. Flowchart of the approach taken to develOp a replacement system for log grading . . . . . . . . . . . . . . 50 15. Frequency distribution of sample log diameter . . . . . . . . . . . . . . 53 16. Frequency distribution of sample log length . . . . . . . . . . . . . . . 53 17. Frequency distribution of sample log original grade . . . . . . . . . . . 54 18. Frequency distribution of sample log geographical source . . . . . . . . . . . . . 54 19. A walnut veneer log illustrating such variables as crook, bark thickness, sap thickness and an unsound knot . . . . . . . . 59 20A. Log sector designation for field data collection . . . . . . . . . . . . . . . . . . 63 20B. Log section designation for field data collection . . . . . . . . . . . . . . . . . . 63 21. Minimum clear cutting and clear volume unit . . . . . . . . . . . . . . . . . . . . . 64 22. Frequency distribution of sample flitch veneer unit price class . . . . . . . . . . . 66 23. Frequency distribution of sample flitch veneer grade . . . . . . . . . . . . . . . . . 67 24. Clear cutting develOpment for 1/8—log sectors . . . . . . . . . . . . . . . . . . . 72 25. Clear cutting develOpment for l/4—1og sectors . . . . . . . . . . . . . . . . . . . 74 Viii Figure 26. 27. 28. 29. 30. 31. Clear cutting develOpment for l/2—log sectors . . . . . . . . . . . . . . . . Veneer price as a function of annual ring pattern . . . . . . . . . . . . . . Veneer price as a function of annual ring pattern transformed to the fourth power . . . . . . . . . . . . . . . . . An example of light bark distortion defects not sufficiently severe to interrupt clear cuttings . . . . . . . . Illustration of a discriminant function Cumulative frequency distribution of veneer price, study firm vrs. competitor firm 0 O O O O O O O O O O O O O O O O 0 ix Page 75 91 93 99 104 115 Appendix I II III IV VI LIST OF APPENDICES Input variable number, description, measurement unit, code and format of program LOGAN . . . . . . . . . . . Variable classification and measure— ment unit of field data . . . . . . . Magnitude boundaries of ranked variables . . . . . . . . . . . . . . . Output variable number, description, measurement unit, code and format of program LOGAN . . . . . . . . . . . . Variable classification and measure- ment unit after preliminary computa— tions 0 O O O O O O O O C O O O O O O O Flowchart of computer program LOGAN . . Page 125 129 131 134 138 141 CHAPTER I INTRODUCTION Black Walnut (Juglans nigra L.) is unquestionably our finest domestic hardwood. From the time the earliest settlers found it in huge trees up to six feet in diameter and 150 feet tall, walnut has served man in a myriad of uses because of its superior characteristics. The pioneer found the clear wood easily split and very durable, hence he used it for split rail fences as well as cabin logs and later as trim and paneling for his fine homes. Walnut has been the primary wood for gunstocks from the time Of the famous long rifle through World War II be- cause of its strength, shock resistance, hardness, and ability to stay in place. Today the major uses for walnut are for all forms of household and office furniture, novelties, and decorative wall paneling. The use trend of walnut timber has been eXpanding and changing since 1933. The volume of walnut wood used for all products ranged from a low of 28 million board feet in 2 1933 to 94 million board feet1 in 1960. Sawmills, veneer mills, and log exporters purchase walnut timber, but until 1960 at least 75 percent Of all walnut logs cut were manu- factured into lumber. However, the volume of walnut wood used in domestic veneer manufacture has doubled in recent years. In 1963, domestic veneer mills used approximately 32 million board feet, international l/4-inch log rule (1). Walnut logs, bolts and hewn timbers are being eXported in increasing quantities. Only 15 years ago less than a million board feet were eXported annually. By 1963, ex— ports had grown to more than 16 million board feet. A de— crease to 11.2 million in eXports in 1964 was the result of a quota placed in effect on February 14, 1964. The control on exports was removed in February 1965, and that year 23.6 million board feet were exported. Exports for the year 1970, the last year of available figures, were 17.4 million board feet (1). This high demand accompanied by a concurrent shortage of walnut timber has set the price structure for this species at a very high level. An average price per thou- sand board feet, Doyle Log Scale, has been reported as high 1A board foot is a volume of wood equal to a board one inch thick 12 inches square. as $1,367 for prime veneer logs 24 inches in diameter and larger. In contrast, the price for yellow—pOplar (Liriodendron tulipifera L.), a nondecorative species, is reported at $171 for the same class and unit volume of material (2). These relatively high unit prices and the lack of adequate methods for quality determination present a most difficult raw material acquisition problem for domestic veneer producers. CHAPTER II THE TIMBER QUALITY CLASSIFICATION PROBLEM A. The General Problem An industry that processes raw material such as mineral ore, basic chemicals or agricultural crops must have basic yield information on the quality and quantity of their specific raw material before purchase and manufacture to make a reliable estimate of the cost of the products to be produced. Success depends in part on a thorough knowledge of the raw material being used. The nature of any firm which converts timber into solid wood commodity-type products is unique because quality and quantity yield of the raw material — the log — are not easily determined in advance, since quality and quantity—reducing defects are often hidden. Unless the output of a sawmill is all 1-inch lumber, the conventional log rules may not provide adequate estimates of product volume. For example, the Doyle Log Rule, because Of blunders in its original con- struction, scales only logs near 26 inches in diameter correctly (3). Classification of timber by quality requires log grad— ing rules, which should be designed to provide (4): l. A reasonably accurate estimate of the quantity and value of product output. 2. A basis for sorting trees or logs to make a particular product or products. 3. Significant differences in value or product yield between grades. 4. Grade specifications that are easily under— stood and applied. 5. Specifications that define the poorest log or tree that may be admitted in the system. 6. Yield data that are based on representative conversion systems. 7. Grades that will have wide application for a tree species throughout its commercial range. 8. Grades that are stable over time. Currently there are a number of grading systems used to determine log quality. These grading rules include: the Purdue Hardwood Log Grades, the Ohio Standard Saw Log Grades, and the U.S. Forest Products Laboratory's Hardwood Log Grades for Standard Lumber. Figure 1 illustrates a typical log— grading specification adapted from the U.S. Forest Service, "Hardwood Log Grades for Standard Lumber" (5). In addition, many companies have modified one of the above rules or have develOped their own quality classes. Figure 1. Forest Service standard specifications for hardwood factory lumber logs. Source: (5) Log Grades Grading Factors F1 F2 F3 Position in tree Butts Butts G Butts & uppers Butts 8 on I} uppe rs uppe rs Diameter, scaling, 113-1s 16-19 20+ 2.11 12+ 8+ inches Length without trim, feet 10+ 10+ 8-9 10-11 12+ 8+ Length, Clear min., ft. 7 5 3 3 3 3 3 2 cuttingé. Number, No on each maximum 2 2 2 2 2 2 3 limit 3 best Fraction faces of log length 5/6 5/6 5/6 2/3 3/4 2/3 2/3 1/2 required in clear cuttin For logs with less Sweep and than 1/4 15% 30% 50% crook of end in allowance sound de- (maximum) fects in percent For logs gross with more volume than 1/4 10% 20% 35% of end in sound de- fects Thtal scaling deduction 5 including sweep and crook 340% —50% 50% l.Ash and basswood butts can be 12 inches if otherwise meeting require- ments fbr small No. 1's. 3.Ten-inch logs of all species can be No. 2 if otherwise meeting require- ments for small No. 1's. é-A clear cutting is a portion of a face free of defects, extending the width of the face. 2.0therwise No. l logs with 41-60% deductions can be No. 2. §.Otherwise No. 2 logs with 51-60% deductions can be No. 3. These rules vary widely as to number of quality classes, quality factors considered, diameter limitations of classes, and the percentage of clear log area specified. In general, great emphasis is placed on diameter limits and the clear area of the log surface expressed as a percentage of the total surface area. The number of faces considered for the percentage clear specifications also varies between grading rules. Some specify consideration of only three faces of the log, while others specify four faces. Another important consideration for domestic veneer and lumber logs is that most grading systems require the complete log be assigned one grade. This aspect will require some analysis later in this study, for each half—log is considered a separate processing entity in the system under analysis. NeWport, et a1. (6) report that in an evaluation of more than 50 known grading systems, with possibly one or two exceptions, all of the systems examined had serious shortcomings. Many of these shortcomings stem from the fact that the grading specifications were determined rather arbitrarily on the basis of estimates of what logs with certain characteristics could yield in a given product. Many of these specifications rely too heavily on the grader's judgement. Performance data may be lacking or sometimes nonexistent. It should be pointed out that many of these current grading systems have some desirable features and sound specifications. The "Hardwood Log Grades for Standard Lumber" of Figure 1 is well accepted by many as an excellent system. The specifications for these grades are correlated closely with the specifications for standard hardwood lumber grades. Hardwood factory lumber is graded on the basis of clear—faced or sound cuttings of a minimum size to comprise a certain fraction of the area of the board. To apply these log grades the log is divided into four faces. Each face is graded as though it were a board, except that rip cuttings and sound cuttings are not allowed. The face must furnish clear cuttings of a definite minimum size to comprise a specified fraction of the face. It has also been determined that the most accurate results were Obtained by using only the three best log faces. This direct correlation of the standard log grades with well defined standards or grades of lumber as a measure of product yield is important. Not only must the final product specification be deterministic and easily communicated, the end product must have specific market values. The hardwood log grading system for standard lumber meets these require— ments. In contrast to the generally well defined hardwood lumber grading rules existing hardwood veneer grading rules are not specific and leave much to the judgement of the grader. Table l is an example of a plain—sliced walnut veneer grading specification. These grades do not specify the number and size of defects allowed in the various grades. Hardwood veneer grading systems vary with the individual plant, the grader and plant procedures. Such systems may not have the consistent relationships nor eXpress a specific market value. Value of the specific grade may vary between markets and over time. Precise specification of product yield in the develOpment of veneer—log grades may be a definite limitation. The U.S. Forest Service has develOped interim veneer— log specifications to aid in timber sale appraisal work (5). To quote M. D. Ostrander: These specifications are based on limited infor— mation Of factors influencing veneer—log quality... They should not be considered final but are in- tended merely to bridge the gap between the present and such time as studies are completed to supply the data for develOping Forest Service standard veneer-log grades and accompanying veneer—grade yield tables. B. Veneer Log Evaluation A graphic analysis of the wood quality—log cost assess- nmnit problem is presented in Figure 2. The required 10 Table 1. Summary of veneer characteristics and defects of premium grade and good grade plain sliced walnut veneer. Characteristics Sapwood Heartwood Color streaks or spots Color variation Mineral streaks Small burls Pin knots Knots (other than pin knots) Worm holes Open splits or joints shake or doze Rough cut Cross bars Inconspicuous patches Type of matching a — Book matched. Premium 10% Yes Slight Yes Slight Yes Occasional No NO NO Source: (7) 99.03 20% Yes Slight Matched for color or grain at the joints (can be slip matched if customer specifies) b — Sharp contrast will not be permitted. 11 Figure 2. Wood quality and the log cost assessment problem. 12 + _ _ m34<> TI .10 ZO_._. L L o mmSowm ZO_._.<._wmmOo Powmzo > 4‘ ommsomm zofigimm wasa> ZO_._. Homma O._. mm_mmzoo 43.911 _ A L m:1_<> to zofiszimm mi mom zowtzimo _ mo_.rm_mm._.o omeEOo _ Axum—23:. .Hmdewvmme whazmwhnfi. 20m... ZO_._._._.Mn:>_Oo l3 correlation is between log cost and the value of the veneer produced from the logs. However, real value of the material is determined only after the conversion process. Therefore, any value or cost assignment of the logs must be an estimate based on the correlation between physical characteristics of the log and the physical characteristics used to determine the value of the resulting veneer. This is illustrated by the circuitous route of value determination (the broken line) in Figure 2. Impinging upon this evaluation is competition from alternate use of the raw material. In this case it would be export for conversion to veneer or domestic conver— sion to lumber. Some illustrations of the effects log surface defects or characteristics have on resultant wood quality will aid in understanding the approach taken in this study. A knot and adventitious bud cluster are pictured as they appear on the surface of a walnut log in Figure 3. How these defect indicators affect the resulting veneer is illustrated in the photograph of the veneer produced from this section of the log is shown in Figure 4. These characteristics reduce the utility of the veneer for certain high quality products. The small holes in the bark of the log shown in Figure 5 are caused by a class of birds called sap—suckers. They 14 Figure 3. Adventitious bud cluster and knot as defect indicators on a log surface. 15 .a find}! I A , .. .4rl. .. \ 15‘,” . A. let? “A?! ,. H .t .. 1....1 at u, may,” 16 _Figure 4. Adventitious bud cluster and knot defects in walnut veneer. l7 . 1. L: mil 18 Figure 5. Bird peck holes in log bark surface. l9 20 feed on the inner bark, cambium and sap of living trees. Many of these holes result in damage to the wood caused by pathological stain entering the tree at this point. The appearance of this defect, termed bird peck in this study, on the end of the log is shown in Figure 6. A consideration in the determination of wood quality is that we are dealing with biological material that is heter— ogeneous. Normally logs and trees vary significantly in their physical, mechanical and chemical prOperties. This variation is Obviously reflected in product yields and value. Figure 7 illustrates the variation in value experi— enced in a stratified sample of log data gathered for this study. There are two important points evident in this data. First, the range in price of veneer from the same grade of log ranges from two cents per square foot to a high Of six and one—half cents. Second, although unit values of timber products normally increase with an increase in diameter class, the only trend evident in this data is almost uniform average value over the range of diameter classes. Conse— quently, data behavior such as this only complicates problem analysis. What may be needed is a store of fundamental informa— tion provided by research such as the recent work by 21 Figure 6. Bird peck as color distortion on the end of a log. 22 23 .mHmEmm xuflamso mOH pmflmflumuum m uOm umumEme OOH mo COADOOSM m we msam> pawflm ummcm> .n musmfim Among: awhmzflo 00.. t m_ m_ S m. N. _ _ p _ Ir 4 _ _ LI nu 24w: mm<1_o mmemzflo Xfi . . u3<> 03-54: . IN 0 o o 00 1m ) nu x m. oo oo- 0.. core. coo I X m... I w m . m m x w u. m I b o oo o oo o o I, w+ o O o 00 I'm . . 7m BOWVA 013M 833N3A 24 Stayton, C. L. et al. (8). They have develOped precise knowledge of the relationship between tree surface abnor- malities and their associated interior defects. To learn more about this variation for sugar maple (Acer saccharum, Marsh.), they determined the percentage Of each type of exterior defect indicator that had an associated interior defect, and observed how this percentage varied with the size of the indicator, its height above the stumps, tree diameter, tree age, and tree growth rate. This and other relations such as the correlation of diameter to quality illustrate some of the complexities of identifying and evaluating quality in logs and trees. Diameter is interrelated with other characteristics such as tree age and growth rate, and the effect of these relation— ships on quality has not been well established. Log length is another attribute of size that has been shown to affect quality. Lane (9) has stated the need for devising more adequate statistical techniques for develOping and testing the per- formance of grading systems. Particularly needed, are electronic computer programs for machine processing of timber quality data. Gaines (10, 11) indicates that the real challenge in log and tree quality work is the develOp— ment of more complete models. We need better eXpressions of 25 the relations between log-surface characteristics and pro— duct yields, and between production—process activities and product yield. In addition to the fact that quality and quantity yields of the raw material are not easily determined, the value added by the primary wood conversion process is low, hence, raw material cost is a high percentage of total cost. The inherent nature of such processes allows for little margin of error. If poor buying decisions are made there is little chance of recouping the loss in the conversion process. CHAPTER III THE DOMESTIC VENEER LOG GRADING PROBLEM A. Study Firm The raw material acquisition problems facing domestic walnut veneer producers became apparent through personal contact with a specific firm. It became obvious there was a general lack of confidence by management and log buyers in their current domestic log grading system. It was decided to study this problem in depth by gathering and analyzing data and Operational information from this walnut veneer- producing firm, subsequently referred to as the "study firm". This firm, located in north central Kentucky, has been purchasing exclusively walnut logs for its sawmill and veneer—log eXport Operations for many years. In addition to lumber, the sawmill has produced half-log flitches, the basic form of raw material for sale to domestic walnut veneer producers. In an effort to grow and diversify, a veneer mill became Operational about two years before this study was initiated. 26 27 The conversion process and material handling system are efficient by current production standards. The actual process is described in chapter four of this study. Annual production is about 20 million square feet of sliced veneer and is equivalent to a sales volume of about one million dollars annually. One unique feature of the Operation is the data proc— essing system. Each individual log is identified from the time it is purchased in the field until it is sold as veneer. This system greatly facilitated the gathering and compila— tion Of the study data. In addition, this data processing system serves as an efficient information system for manage— ment. Log buying efficiency can be assessed by individual buyer and by groups of logs. Log value can be assessed by geographical area of origin. Most important, log grade specifications can be tested for their ability to prOperly differentiate quality levels (value) of resultant products. 1. General Quality Levels Currently the study firm, and other firms in the in— dustry, believe that profit from purchasing and processing walnut logs can be maximized by segregating logs into three distinct quality levels. Premium quality logs, the very choice, as illustrated in Figure 8, are eXported to EurOpe 28 Figure 8. Premium quality walnut logs such as this are exported to EurOpe for furniture veneer production. 29 . . vi; a 4|ALnNr.C,.uv- no .1 315.»! .. r . , . . .c.. .I.\b. 3‘0. .. . 3‘ . v 4.; 30 for conversion into veneer for furniture manufacture. Medium quality logs are processed in this country, gener— ally into flat—sliced veneer for furniture and wall panel— ing products. An example of this quality level of material is shown in Figure 9. Logs judged not capable of being pro— cessed profitably into veneer are sawn into lumber. This lowest quality class of logs is illustrated in Figure 10. These three general quality levels can be compared to the three broad log—use classes outlined by the United States Forest Service in their approach to hardwood lOg grading. They each cover current utilization practices. The Forest Service's division by use is as follows: (1) factory class, the three grades as described in Figure l, (2) construction class, an ungraded category intended to be used for structural purposes, (3) local—use class, those suitable for products not usually covered by standard pro— duct Specifications. 2. .Domestic Veneer Log Grades The normal procedure for a sawmill buying woods—run logs of a species other than walnut, and using the Forest Services log grades would be to have four grade sorts: factory log grades, one, two, three, and a local—use or industrial grade. If veneer logs were sold, they would be A... 34...? |||I|J1n 31 Figure 9. Medium quality logs as illustrated here are processed domestically into flat—sliced veneer. 32 Figure 10. 33 Lower quality walnut logs such as these are sawn into lumber. 35 withdrawn from factory grades one and two. This procedure is somewhat different than the operating procedure of the study firm. They have taken woods—run logs and divided them into three general quality levels previously described. These general quality levels have been further divided as follows: 1. EXport veneer logs—grades 1 thru 6 2. Domestic veneer logs—grades A, B, C, D, E and F 3. Domestic lumber logs—grades Ll thru L4 The grade description and instructions for use of the domes— tic veneer logs are presented in Figure 11. Not only are these grade rules complex, they are changed from time to time without verification from deter- ministic yield data. Information such as that contained in Figure 7 illustrates that the application of these quality level specifications by log buyers in the field does not enable them to satisfactorily separate logs into quality levels to which distinct average end—product values can be assigned. This ability is required to allow buyers to determine what maximum price can be paid for a given quality of logs and still allow the firm to make a reasonable profit with a minimum of risk. This is one of the basic problems facing the firm and one to which this study has been directed. 36 Figure 11. Domestic walnut veneer log grade rules. Instructions: Divide veneer logs into two faces corresponding to flitching line. Measure off all unsound defects and double heart befbre determining length, diameter and grade. Measure diameter inside the bark at the smallest circumference (not at swell where tree has started to fork or branch). Drop second cut logs one grade, as grades are fOr butt logs only. Do not cut back log length on tally for purpose of raising diameter - logs must be bucked. A Fresh cut, woods grown, and judged practically free of pin knots. 9' to 16' in length (normally 13' maximum). Diameters 16" and Up, except allowing 15" logs with 13" of sapwood. Measured diameter must be 90% heartwood. Logs containing 81' to 134' must be free of all defects; 135' and Up must yield 95% in two cuttings. Requires excellent color and conformation, no worms. Minimum cutting length is 8'. B Fresh cut, woods grown, judged reasonably free of pin knots. 8'-6" to 16' (normally, 13' maximum). Diameters 15" and Up, except will allow 14" log with 12 1/2" of heartwood if 9' or longer. Measured diameter must be 85% heartwood. Logs 60' to 80' must be free of defects; 81' to 105' must yield 95% in two cuts; 106' and Up, 90% in two cuts or 95% in three cuts. Requires very good color and no signs of worms or peck. If logs shorter than 8'-6", drop to C. Minimum cutting is 5'. C Logs 6' to 16', 14" and Up, except 6' G 7' logs must be at least 15". 80% of measured diameter must be heartwood. 45' to 59', free of defects; 60' to 80', 95% in two cuts; 81' to 105', 90% in two cuts or 95% in three cuts; 106' and Up, 85% in two cuts or 90% in not over four cuts. Good color, minor signs of worms allowed only if logs 100% clear, 9' — 13' in length. Allow no more than 20% 6' and 7' logs in any purchase, dropping balance to grade D. Clear logs, good color, no worms, 9' to 13', 14" and larger, having metal in one face, may be graded C, with measure- ment cut in half. Minimum cutting 5'. D Same as C, except admits 45' to 59' yielding 90% in two cuts; 60' to 105' yielding 75% in two cuts, 106' and Up yielding 75% in not over four cuts. Drop excess (over 20%) 6' and 7' logs one grade. E Same as C, except admits 45' to 105' yielding 50% in one cut; 106' and Up yielding 50% in not over four cuts. Excess over 20% 6' G 7' logs to be placed in appropriate lumber grade. F Butts only, 9' to 13', 12" and 13", thin sapped, quality and conformation equivalent to Grade A. 37 Table 2 illustrates the results of the application of the six log grades described in Figure 11 for an Operating period of one month. Adjusted gross income, line one, is the total sales value of the resulting veneer per thousand board feet(MBF) Doyle Log Scale less all costs except direct log costs associated with each log grade. Quality level for these grades would generally be eXpected to decrease from left to right. The actual values, however, indicate that the tOp three grades (A, B, F) yield veneer priced at nearly the same level. The lower three grades (C, D, E), while somewhat more variable, yield veneer of a distinctly lower level of value. Log cost, line two, reflects the general trend of assumed value level of log grades. These costs reflect the competitive situation of various buyers and quality aspects of geographic area of purchase. Net pre—tax profit, line three, is the profit or loss in dollars resulting from purchases of individual log grades for the month. The range in pre-tax profit is extreme. It varies from a loss of $285 per MBF for grade A logs to a profit of $397 for grade F logs. Twenty percent of gross income has been set by the firm as a net pre—tax profit goal. The variance in dollars from the stated profit goal by log grade appears on line five. 38 Rome- Reno- Rmefi+ Rmmml Rema+ Rana: Roma- 8 Aucmoumav pmchuum Hmom DAMOHQ mo cofiuuom Rm- Hm- _ mHH+ Ham- mea+ Nae- Ham- m I44 mafia mme m mcflav wocmflnm> meg Hes sea NMH 4mm Hem mam a Auauoum mmoue Roms Hmom bemoan xmbumum umz om oHH Ham mean Ram HRH- mmmu m ufluoum xmuuwum uwz new mmm mmm Rmm mum mums «Hes N umoo mos mom mos mam mmo News 40mg HMHH H msoocH mmOuO Uwumshoa ammum>< m a O m m a mafia EmbH mm0 BDZQGB UHBmmEOQ zmHm Aoamum moq mahoa poem mwomum mOH uwmcm> OHumeOo now uuomwu moaneum> uflmoum hanucoz oumon oooa Mom meHHOQV .m mHQmB 39 The portion of profit goal attained, line six, is the net pre-tax profit goal expressed as a percentage. This ratio is used as a management tool to measure the effective— ness of the log purchasing system. An analysis of the ratio and the cost—value variance leads one to pose such questions as, "Why is it necessary to pay unprofitably high prices for log grades A, B and C while grades F, D and E are more in line with company objectives"? These results, which are typical, reflect a very inefficient grading system, in fact, money and effort Spend on such a system may well be wasted. This and other questions leads us to the Objective of this study. B. Objective The objective is to critically analyze the domestic walnut veneer-log purchasing, flat—sliced processing and veneer marketing system within a single producing firm. The purpose is to search for methods by which improved efficien- cy of the system may result. A second Objective is to test the use of modern analytical procedures and statistical methods in an effort to predict the unit value of resulting veneer from the exterior characteristic of the log. Any results which would increase the efficiency of this log— purchasing system or the system in general would be a ~— v—_‘:-' 'u _,—F 40 significant contribution to this important phase of forest products marketing and production. CHAPTER IV THE STUDY FIRM'S ACQUISITION, PROCESSING AND MARKETING SYSTEM A. Acquisition The organization of the study firm's operating system does not vary significantly from the typical walnut veneer firm. The following chart of Figure 12 illustrates this system. The acquisition or buyer organization does differ from the typical hardwood veneer mill because of the scat— tered nature and geographical range of walnut. Eight to ten buyers operate in the walnut—processing area of Ohio, Indiana, Illinois, Kentucky and Tennessee. With only one supervisor to oversee these scattered peOple, it becomes obvious they operate with a minimum of supervision. The buyers are also responsible for logging timber, concentra— tion and shipment of purchased logs to the central proces— sing plant. Generally buyers are not professional foresters but have a wide range of educational backgrounds. Most have working experience as loggers or related Occupations. Many have had previous experience buying walnut or other species 41 42 Figure 12. Flowchart of study firm's operating system. 43 FIELD LOG 8I TIMBER PURCHASING I RECEIVING-GRADE SORTING AT CENTRAL PROCESSING I i I EXPORT DOMESTIC DOMESTIC VENEER—LOG VENEER LUMBER SALES PROCESSING PROCESSINGfi HALF - LOG CONVERSIONT FLITCHES TO LUMBER 4 CONVERSON TO VENEER ‘ SAMPLESIVENEER GRADING AND PRICING I * WAREHOUSE I STORAGE MARKETING 7 AND SALES :4 SHIPMENT *— 44 of timber or timber products. The mode of independent Operation and the large sums Of money involved can well in— fluence the temptation to Operate for the buyers own well— being. Authenticated and hearsay accounts of timber and log thefts, fraud and kickback are often heard in the walnut timber business. An informal training period of up to six months is given each new buyer, depending on his previous training and eXperience. Log-grading, company Operating procedures, and pricing policies are stressed in training. New buyers work under the supervision of an eXperienced buyer; first in the receiving yard and then in the field. They are then assigned a buying territory of their own. B. Processing After the logs are purchased and shipped to the central processing site, important receiving, inspection and grading functions are performed. Several very important decisions are made at this point. They are: 1. Final log grade assignment 2. Allocation to: a) eXport sales b) domestic veneer processing or, c) lumber processing 45 3. Determination of flitch orientation by establishing central saw line on end sections of domestic veneer logs only. The conversion of half—log flitches is performed by conventional sawmill equipment. Logs are slabbed and sawn in half as illustrated in Figure 13. This is a critical processing decision center, for once the central cut is made the location of defects is determined in the resulting veneer flitch. Company policy calls for the sawyer to saw the flitches as prescribed by the receiving inspector. The flitch line marks are evident on the end of veneer logs as in Figure 9. However, as the process continues there are two pro— cessing steps at which these decisions may be altered. First, after the logs are debarked; second, after slabs are removed from at least three of the log's faces but before the flitch line is sawn. Each of these actions may reveal information which could alter the decision made by the in— spector. Under these conditions the sawyer is allowed to change the decisions made by the receiving yard inspector. Domestic lumber conversion on a production basis is performed by the same personnel and conversion unit that is also engaged in veneer flitch production. The decision process and production rate for lumber production varies ' \ A l I I I I SLABS 47 considerably from that required for veneer flitch conver- sion. The conversion of half—log flitches at the veneer plant includes heating prior to slicing, slicing, steaming (aging) before drying, drying, and packaging prior to movement to the warehouse for storage. Veneer area, in square feet, Of each flitch is determined automatically at the outfeed end of the drier. Three "sample" sheets of veneer from the thickness quarter-points of the flitch are also removed at this point in processing. These sample sheets represent the quality of the complete flitch. They are the basis for the assignment of product, grade and unit price of the flitch. After these assignments are made the sample sheets are used by the marketing division as a sales tool to represent the quality of the entire flitch to potential buyers. C. Marketing The two general product classifications of walnut veneer are furniture manufacture and decorative wall paneling. The use requirements for veneer in these products may differ con- siderably. Within the paneling product class there are three sub—classes: (l) architectural (2) commerical and (3) architectural door skins. A length of at least 100 inches is a requirement for veneer used in panel manufacture; 48 except for door skins. Size of major defects and frequency of minor defects are important factors by which grades are segregated within this product class, grain pattern and color are very important quality factors for better panel grades. Furniture manufacture requires clear cuttings of a specific length for a specific end use. Long length and a high percent yield are very important quality factors. Grain pattern and color within the clear cuttings are other important determinates of quality within this product class. Matching of grain pattern is especially important in higher quality furniture production. Minor defect frequency, grain pattern, color and color matching of veneer are less important in the manufacture of lower quality furniture. CHAPTER V PROCEDURE, SAMPLING AND DATA COLLECTION A. General It was obvious from the analysis of the current log grading system in use by the study firm that it does not meet log value estimation requirements. The system does not adequately separate logs or groups of logs into quality classes with significant differences in value of end pro— ducts. One objective of this study was to use modern quan- titative and statistical methods to develOp a replacement system: a system which will predict the unit value of veneer from a quantification of the exterior characteristics of the log. The procedure used for achieving this objec— tive is outlined in Figure 14. B. Procedures In general, the approach was to: carefully select the study material (veneer logs) over a range Of physical sizes, quality levels and geographical ranges; collect the neces— sary raw material, conversion process and final product data; convert the data to useable form; develOp a predictive 49 50 SAMPLING DESIGN I DATA COLLECTION COMPUTATION OF FIELD DATA STATISTICAL EVALUATION I. Regression analysis 2. Discriminant function analysis ANALYSIS OF RESULTS CONCLUSIONS Figure 14. Flowchart of the approach taken to develOp a replacement system for log grading. 51 statistically sound model; analyze the results and state the conclusions. The problem was to establish numerical expressions for each log characteristic (independent variables), which were then related mathematically, using multivariate analysis and electronic computers, to an expression of product quality (dependent variables). Many characteristics must be tested to evaluate their contribution toward predicting the value of the resultant product. The complexity of this problem can not be overstated because Of the large number of variables and the interrelationship that exist. Some of these characteristics are visible defect indicators on the log surface, some have to do with tree form, and others are in the interior with no obvious outer evidence of their existence. C. Sampling A determined effort was made to stratify sampling over the range of log sizes and surface clearnesses available. Since quality variation in the pOpulation was unknown an estimate of sample size could not be made. Henley et a1. (12) state that if past experience with lumber logs is indicative, a sample of 30 logs per one—inch diameter class would be required for each species if they are representa— tive of the possible quality range. 52 The objective in log size sampling was to select 30 flitches (15 logs) in each one—inch diameter class between 12 inches and 24 inches. The skewed distribution of Figure 15 illustrates the actual sample of the diameter classes obtained. Although it was generally agreed that logs under 12 inches were not to be purchased their Occurence was such that 20 flitches were included in the sample. Domestic veneer grade logs with diameter class greater than 15 inches were not sufficiently available to meet the above require— ments during the five—week period data was collected. It is believed the length distribution of the sample as illustrated in Figure 16 represents quite well the actual population. The maximum length capable of being processed is limited to 13.5 feet by the slicer. The small number of logs shorter than eight feet reflects the limitation that short logs are inefficient to process and company log grades discriminate against them. The histogram of Figure 17 represents the quality dis— tribution of the sample flitches as expressed by the current log—grading system of the study firm. Logs of grades one through six are normally shipped as export logs. Grades A through E are domestic veneer log grades, and grades L1 and L2 are the tOp two of four lumber log grades. The current 53 —_I - if N-282 E a "—7 ”—1 Q It“ __. K ,._.4 ‘ I———-.____ I0 II I2 I3 I4 l5 l6 I7 I8 l9 20+ DIAMETER CLASS (inches) Figure 15. Frequency distribution of sample log diameter. I — N=282 FREQUENCY > 56 7 8 9 I0 II I2 I3 I4 LENGTH CLASS (feet) Figure 16. Frequency distribution of sample log length. 54 I '7" I I VENEER ' N"- 282 : LOGS (208) I XII I y I 9 I _‘ I a ' I LUMBER § EXPORT I —I LOGSI29I 8 L068 (43) I r— __ I Pi ' F" I I I I I2 3456iABFCDEILIL2 NUMERICAL LOG GRADE Figure 17. Frequency distribution Of sample log original grade. I63 ,_.__T N = 282 XA 3 L... 87 S ‘——1 o 3: LL 9 5 IO ['"1 r—-I I—_I TENN. KY ILL. IND. OHIO GEOGRAPHICAL LOG SOURCE BY STATE Figure 18. Frequency distribution of sample log geographical source. 55 domestic and lumber grades were assumed to represent a stratified sample of quality. Export grades were included to gain additional sample space for diameter classes greater than 15 inches. There has been some evidence that geographical source as a quality indicator is important in wood quality studies. An effort was made to include logs from the total area over which logs were normally purchased. The sample frequency distribution by state is shown in Figure 18. It reflects what is considered to be the long term purchases by area of the firm. D. Data Collection Log and veneer information was Obtained by a procedure specifically develOped by Bulgrin (13) for wood quality evaluation studies. This involved diagrammingl the sample logs, following them through all stages of processing and recording pertinent information including the resultant quality of each veneer flitch produced. The exterior surface of every log was examined, defect indicators, and all identifying data recorded. All surface abnormalities were considered potential defects, with no IA log diagram is a graphic representation of the exte- rior surface and end section Of a log that accurately depicts the nature and location of observable and definite features that are known to or may affect the utilization of the log. 56 preconceived ideas of their significances. These data on defect types, size and locations provided a complete de— scription of the exterior surfaces of the logs. For purposes of this study the variables, both inde— pendent and dependent, have been classed as follows: 1. Measured variables a. Continuous b. Discontinuous 2. Ranked variables 3. Attributes 4. Computed variables See Appendix I for full documentation of the variables, their identification and the coding methods used for elec— tronic computer processing. 1. Measured Variables a. Continuous Measured Variables. These variables can be expressed in a numerically ordered fashion and can assume an infinite number of values between any two fixed point values. The last digit Of the measurement should imply precision, that is, the limits on the measurement scale between which we believe the true measurement to be. The continuous variables recorded for the logs included in the study are listed in Appendix II. They are variables such as length, diameter, crook and sweep; variables one would 57 expect in such an analysis. The log in Figure 19 illustrates many of these variables. Crook is, of course, the most evident characteristic of this log. These variables normal— ly present no problems of eXpression or interpretation and are most easily handled in statistical and other analytical procedures. b. Discontinuous Measured Variables. These are varia— bles which have only certain fixed numerical values, with no intermediate values possible between them. A limited number of variables in this study are discrete. In fact the only one in the list of raw data is the number of log surface defects. 2. Ranked Variables Some variables cannot be measured but can at least be ordered or ranked by their magnitude. The difference in magnitude between ranks is not necessarily identical or even prOportional. Color, for example, is a very important factor in the determination of walnut veneer quality. There are SOphisticated methods of color measurement and communica— tion. These methods present problems when an attempt is made to apply them to wood (14). As a result of these problems, heartwood color designation at the end of fresh cut logs was assigned one of the color ranking described in Appendix III. Figure 19. 58 A walnut veneer log illustrating such variables as crook, bark thickness, sap thickness and an unsound knot. 59 60 Other variables such as stain, grain contrast, ring pattern and freshness were assigned similar ranking. 3. Attributes Variables which cannot be measured but which must be expressed qualitatively are called attributes (Examples — black, white; male, female). When such attributes are com— bined with frequencies they can be treated statistically. Certain attributes that can be ranked or ordered can be coded to become ranked variables. The attributes recorded for this study are listed in Appendix III. Log type is an excellent example of this class of variables. The butt, or bottom, log of a tree is generally considered to contain the highest quality wood because of the general lack of limbs or knots which results in defect—free wood. Any logs above this first log are a second class and are referred to as upper logs. 4. Computed Variables The majority of variables in any biological work such as this are observations recorded as direct measurements, counts or perhaps outputs of instruments. However, there is an important class of variables which we may call computed or derived variables. These are generally based on two or more independently measured variables whose relations are 61 expressed in a certain way. Ratios and percentages are examples of this class of variables. A ratio expresses as a single value the relation which two variables have one to the other. The number of pin knots per square foot is an example of this type of variable. The two independent variables are (1) number of pin knots observed on the surface of a log (2) the area of the log which in turn is the product of the circumference and length of the log. The resultant computed variables is the Occurance of pin knots per unit area, in this case, per square foot. D. Defect Recording System Most log quality specifications require certain frac— tions Of the log surface to be clear of defect indicators. The system developed for use in this study to manipulate input data to calculate various combinations of clear cutting length and width is unique. The following is a description of that system. Certain physical divisions were defined and used to record the location of visible defect indicators such as overgrown knots, sound and unsound knots, seams, bumps, surface rises, and bark distortions on the bark surface and end of the log. The end of each log was divided into eight tfi" 62 sectors as illustrated in Figure 20A. The location of the flitch line as prescribed by the receiving log inspector served as the reference line for sector location. The log was further divided into six-inch sections along its length (see Figure 20B). This develOpment is a grid or matrix system which will allow location of defects by coordinates. Defects can easily be located by a sector (face number) and a section number. For example, the top knot in Figure 20B would be located in sector one, section 11 and 12, the other knot in sector four, section 17. The basic unit for calculations is a pie—shaped cylinder section with dimensions as shown in Figure 21. To more efficiently express the effect of clear wood volume on quality this unit was divided in half on a volume basis. The dark area is the outer volume layer, the light area the inner volume layer. This division allowed specification of defects into three severity levels as follOws: Level 1 — If the defect affects only the outer volume layer Level 2 - If the defect affects only the inner volume layer Level 3 — If the defect affects the total volume (both volume layers) The defect indicators were recorded with these designa- tions along with other log information on the diagramming form. This system allowed the computation of clear areas 63 FLITCH LINE Figure 20A. Log sector designation for field data collection. A T 1 MADE A I I2 345 6789IOIII2|3I4I5I6I7|8I920 Figure ZOB. Log section designation for field data collection. .DHCO OESHO> HOOHO tam mcfluuso HOOHU ESEHCHS mmroz_ x_m .HN Ousmflm 65 of log surface and volumes of clear wood by specification of cutting width in number of log sections and minimum cutting length. F. Yield Data Several product yield variables were determined. Flitch length was recorded as a check against log length. Flitch area, in square feet of veneer 1/35 of an inch thick, was determined mechanically as a normal firm Operating procedure. Flitch area multiplied by the unit value determined the price of the resultant veneer. The nearly normal frequency dis— tribution of unit veneer value is illustrated in Figure 22. The mean unit value was 4.2 cents with a range from 1.0 cents to 8.0 cents. In addition to price, the veneer produced was assigned a grade designation. This assignment was governed by quality characteristics of the veneer for use in furniture or wall paneling production. The characteristics for these individual products are described on page 47. The composite product frequency distribution of these grades is shown in Figure 23. Seven of the grades are furniture veneer pro— ducts; six are paneling veneer products. In general the quality level decreases from left to right. a 66 Figure 22. G _. E 3 A: u. "—‘I I.O 2.0 3.0 4.0 5.0 6.0 7.0 8.0 PRICE CLASS (cents/sqft.) Frequency distribution of sample flitch veneer unit price class. 67 PANELING GRADES [:3 FURNITURE GRADES x’I‘ E A 5E K l2 34 5 67 8 9IOII|2I3 NUMERICAL VENEER GRADE Figure 23. Frequency distribution of sample flitch veneer grade. CHAPTER VI PRELIMINARY COMPUTATION OF FIELD DATA A. General The previous chapter described the collection and classification of field data. Before these variables were tested for their ability to predict the value of veneer yield in terms of quantity and quality, data conversions were made which would increase their predictive value. A computer program, LOGAN (LOG ANalysis), was written to handle the mass of data and to compute the various clear cutting width and length combinations deemed necessary to test. The computations fell into three classes: (1) log characteris— tics, (2) log volume corrections, (3) clear area cuttings and clear volume computations. B. Log Characteristics There were a host of alternatives available for combina— tion and conversion of the log characteristic variables. Certain decisions were made intuitively while other conver— sions were made and subsequently tested against alternatives to determine their relative merits. 68 6 9 Log sweep was converted to a computed variable - sweep in inches per foot of log length — to reflect the effect of log length. For example, an eight—foot log with three inches of sweep and a 12-foot log with three—inch sweep do not have the same degree of sweep. Mean values (large and small-end averages) of bark thickness, sapwood thickness and pith displacement were calculated and were used as variables rather than individual log—end values. This was done to reduce the number of variables and to more nearly reflect the average condition within the log. The ranked variables Of bark growth visibility, new bark color and old bark vigor were combined into one variable — mean bark vigor. Attribute variables log type and growth rate uniformity were converted to ranked variables. Butt logs were ranked as the higher value log type while upper lOgs were considered lower value logs. For the growth rate uniformity variable, logs with uniform growth rate were ranked above those with non—uniform growth rates. C. LogAVolume Correction Although most walnut logs are purchased using the Doyle Log Scale as a volume determinate, it poorly reflects the actual board foot volume of the log. It underscales small logs and overscales those over 26 inches in diameter. To 7O insure accurate log volume determination for this study, the log was assumed to be a cylinder with a circular cross sec— tion equal to the small—end diameter. The equation used to determine board feet of a flitch was: 2 V = 0.00545416 x D X L x 12 2 = 0.03072472 x O2 x L (4) Where: V = volume — actual board feet D = small end diameter — inches L = length - feet These volume calculations were corrected for several factors to reflect more accurately their actual volume. Reductions were made in flitch volume to correct for log sweep and crook. Volume adjustments were made between flitches of the same log for two reasons. First, pith location is an impor— tant reference point for grain pattern develOpment. The normal practice, if both flitches are to be sliced, is to saw the log into flitches of equal volume using the pith as a guide. Logs with displaced piths are sawn as close to the pith as possible. This action results in flitches whose volumes must be corrected to reflect the amount of pith diSplacement. Second, those logs whose quality and/or size 71 dictate that one half must be processed into lumber, are sawn so that the pith is located approximately one inch into the veneer—flitch half of the log. This practice re— sults in increased yield by allowing the flitch to be sliced to the pith. These volume correlations were used to reflect changes in the basic unit volume for clear cutting calcula— tions. D. Clear Area Cuttings and Clear Wood Volume Computations The most complex problem was develOping a technique for transforming defect location and severity level information into log surface and volume data for various combinations of clear cutting lengths and widths. To keep this develOpment relatively straight forward, the initial assumption is that the defect severity level is three, that is, the basic volume unit is completely defective. The minimum width of clear cutting, one sector (1/8 of log circumference), of a ten—foot half log is illustrated in Figure 24. The two hatched areas represent defective basic units. If we assume a minimum clear cutting length of six feet, the dotted portion represents surface area, hence log volume, EQE qualified for clear cuttings. The light area represents three clear cuttings one sector wide, two of which are ten feet long and the third eight feet long. 72 .wuouowm mOHIm\H Ham OBOEQOHO>O© mcfluuso ummHO .vm wusmflm m "9.30 m fiwm "0003 ""3640 I_<._.O.r Hum: m 801038 Hum“. m u 0275.30 mOp mcfluusu HOOHO .mm wusmflm _ utho m Nmm n 0003 mO© mcfluuso umeO .om wusmflm Ontho m WEBER ow on :0.» _:om ON 0. o _ _ 4 4 <_A___ __<__fi.1 H d < \ 36QO \044 o 0 G ...e A». Go 0.99 A nIU o o I x I] 3 o ,I O 30 I o I o O x lI O o x x / Ill! . no N a— y x cm H 2 a“ x x I. M M 0 H X W [hr-l cm 0 5 I e x x x x Ii W x O x x h 3 x I. NEEDS o A O 4 \ $.5QO .. ,m VARIABLE x2 105 is included with species 2, all other members of species 1 are in a group by themselves. Clearly, the discriminant function has served to separate the two groups. If you are faced with a new, unknown specimen, you would measure X1 and X2 for it, from these calculate a value of Z employing the previously calculated coefficients b1 and b2 and with an estimable degree of accuracy allocate the new individual to the correct group. This is a very useful device whenever we need to 'I identify unknown specimens and assign them to previously recognized groups. This is true in group identification where previous classes have already been established. Where the previous groups have not been established, discriminant function is of no assistance. Statistical tests are avail— able to test the significance of the equations developed and the contribution of individual variables or sets of variables. These tests are very similar to those used in regression methods. 1. Model Testing In our case this technique was used to classify logs into one of two groups in which the value Of the veneer produced is significantly different. Two basic models were developed and tested. In model I veneer price was the basis 106 for assignment to the reCOgnized groups for develOpment of a predictive equation. Veneer with a value of 4.0 cents or more per square foot comprised the higher quality grade, below 4.0 cents, the lower grade. The variables listed in Table 5 were included in the analysis. They were selected on the basis that they were the eight most significant variables in the regression analysis. For model II—A the assignment to the recognized groups was on the basis of the length of the longest clear cutting. Half-logs with a clear cutting of eight feet or longer were selected for the higher grade, between four feet and eight feet for the lower grade. Model II-B was tested using the grade division of model II—A, but included only variables significant in the analysis of model II-A. 2. Results The analysis of the model I resulted in a nonsignificant equation for distinguishing between the two grade classifi- cations. Model II—A resulted in a highly significant equation, the results of which are shown in Table 5. Two of the eight variables were significant at the 10 percent level. The results of model II-B are shown in Table 6. The equation 107 Table 5. Results of the discriminant analysis test of the two grade, eight variable model (model II—A). Variable Discriminant Significant Coefficient (10% level) Pin knots 0.1724 No Sweep —3.l403 No Crook —0.3503 No Length -l.l771 Yes Minimum clear cut length —O.1257 Yes Geographical source -0.2469 NO Ring pattern — fourth power 0.0015 NO Freshness —0.2824 No 108 Table 6. Results of the two grade, two variable discrim- inant model (model II-B). Function: Variable Coefficient Significance Level Length —1.l349 0.001 Minimum clear cut length —0.1230 0.001 Classification Matrix (logs): Grade Grade 1 2 Total Correctly Assigned l 63 3 66 95% 2 7 95 102 93% Mean Value Difference: Grade Value Per Square Foot (cents) l .91 2 4.32 109 for this model was also highly significant. Log length and minimum clear cutting length were the significant variables. The equation correctly assigned 95 percent (63 of 66) Of the sample logs that were grade one to that grade, and 93 percent (95 of 102) Of the sample logs that were grade two to that grade (see Table 6). The mean value difference per square foot of the veneer is 0.41 cents higher in the lower grade of logs. CHAPTER VIII ANALYSIS OF RESULTS AND CONCLUSIONS A. Background One objective of this study was to test the use of modern analytical procedures and statistical methods in an effort to predict the unit values of resulting veneer from the exterior Characteristics of the log. A firm with such predictive capability and a knowledge of current selling, administrative and manufacturing costs could determine maximum raw material cost levels and still maintain a high probability of reaching predetermined profit goals. The purpose Of this effort was to improve the current grading system in use by the study firm. B. Analysis of Results In summary, however, these modern analytical methods have not proven more successful than the present industry methods. Neither system really works but people in industry certainly know walnut timber; how to buy, process and sell the products. This discussion is in no way faulting the 110 111 system, for we have not been able to improve on it. What may have to be done is to improve or completely change the marketing system. The ability of the initial model to account for only 13 percent of variability of the unit price of veneer with 37 variables is of little consequence. At least half (50 percent) of the variability should be explainable with a limited number of variables before a model would begin to be considered successful. Model improvement techniques also failed to significant— ly improve the predictive ability of the regression equation. Four variables accounting for 19.6 percent of the variation of veneer unit cost in the final model cannot be considered satisfactory results. The inability to develop a significant equation for distinguishing between grade classification by discriminant function methods can only be interpreted as failure for this method. This result confirms the low coefficient of multiple determination of the regression analysis. Efforts to separate log grades on the basis of clear area resulted in a signif— icant equation but mean value difference was less than one— half cent per square foot. 112 Although only a small portion of the variability of veneer value can be accounted for, the study has identified those variables making this contribution. They are: mini— mum clear cutting length, log length, geographical log source, and ring pattern. This information should give direction to future research. The system developed for recording data, manipulating this data and the computer pro— gram for determination of the clear cutting variables is considered a significant contribution to timber quality study methodology. C. Conclusions The question is, of course, what factors have been identified as being responsible for failure of the quantita— tive methods and what conclusions have resulted from the analysis of the study firm's purchasing, pr0cessing, and marketing system. The factors and conclusions fall into two general classifications, they are: (1) marketing system limitations (2) buying system limitations. 1. Marketing System Limitations a. Veneer grading. Perhaps the most serious limita— tion of the current walnut veneer marketing system is the lack of Objective veneer grading rules. The current system lacks one Of the basic requirements Of a grading system; that is, the end products must have well—defined standards 113 or grades before a grading system can be developed for logs that produce that end product. This is a general industry limitation and not one applying specifically to the study firm. Grading rules similar to the hardwood lumber grades could be developed. Clear cutting lengths could be speci— fied with defect definitions developed to specify defects which would interrupt the cuttings. Yield in percent clear veneer for each grade could be Specified. This limitation weakens the statistical analysis by introducing error in the dependent variable — price per square foot of the veneer. b. Pricing. There is sufficient evidence to conclude that the study firm's marketing inexperience is seriously limiting its profit potential. Cumulative frequency distribu- tion curves of unit veneer price of the sample data taken from the study firm and inventory data from a competitive firm are shown in Figure 31. There is an obvious difference in these price distributions. It should be noted here that information from the competitor firm became available after the study was completed. The veneer price range for the study firm is 1.0 to 8.0 cents per square foot; while the competitor price range is 3.0 to 60.0 cents. The data also identifies significant differences in price distribution at the higher value levels. The tOp 10 percent of unit veneer Figure 31. 114 Cumulative frequency distribution of veneer price, study firm vrs. compet— itor firm. 115 cm mm £82.83 were mmmzm> mm mm am om G__ N“. m w _ P _ P _ SEC motkwnioo III \ SEC >095 III TOO. CUMULATIVE FREQUENCY (°/o) rt 116 Table 7. A comparison of the differential pricing policies of the study firm and a competitor. (20,000,000 S.F./year production) Study Firm Competitor Quality Class average yearly average yearly price income price income (¢/SF) (dollars) (¢/SF) (dollars) A11 Walnut Veneer 4.10 820,000 6.16 1,231,200 Top 10% of Quality 6.12 122,400 17.29 345,980 117 prices range from 5.8 to 8.0 cents for the study firm; the competitor firm from 8.0 to 60.0 cents per square foot. This evidence identifies possible causes for failure of the regression models. One could not expect the model to explain price variation from independent variables if unit price differentials were present but not identified in the dependent variable. That is, the log characteristics (independent variables) may have been at such levels that 15 cents per square foot veneer (dependent variable) would have been predicted; the veneer may have been of that value, but was incorrectly priced at eight cents. Under these conditions it is Obvious that mathematical models would fail. The effect on yearly income generation of this price differentiation is presented in Table 7. Annual income for the equivalent production (20 MM square feet) by the com— petitor is $411,200 greater than that of the study firm. The raw material acquisition cost is not available for the competitor. These costs can not be expected to vary greatly; for both function in the same purchasing environment. Real income differences are evident when the tOp 10 percent of the price ranges are analyzed. The income increment from this portion of production for the competitor is over 118 $223,000 — more than one—fourth the projected annual income of the study firm. c. Specialized Markets. In addition to imperfect pricing, the study firm lacks general marketing expertise. It has identified very few specialized markets; lacks the experience to identify these veneers; and fails to hold uniquely figured flitches in inventory to meet market Opportunities. For example, a furniture company in North Carolina has designed a bedroom suite with highly figured walnut veneer. This unique grain figure is caused by pin knots. A limited number of pin knots generally result in degrade of otherwise high quality veneer. However, when pin knots appear in excessive numbers highly figured veneer results. Without knowledge of this specialized market, veneer with numerous pin knots may be priced as low quality rather than high quality and may not be inventoried for this specialized market. Value determination of domestic veneer should be the basis for value estimation of domestic veneer logs (Figure 2). Impinging upon this valuation process is, however, competition from eXport veneer markets and lumber conversion markets. The real competition for domestic logs is the ex— port veneer log market. Why is this so? Apparently the EurOpeans Obtain more value from the products of the logs 119 than do domestic producers. It would seem a logical move by the study firm to determine what product or products the EurOpean veneer mills produce and take steps to Obtain a share of this market. Domestic firms would have certain competitive advantages. Transportation costs would be one, ' for water lost in drying and conversion residue would be a significant weight factor. Knowledge of walnut timber sources, supply, quality and market alternatives would be others. d. Supply—Demand Disruptions. There are additional influences of the competition from the eXport veneer log market on domestic veneer log purchasing systems. Until the export market develOped for walnut logs the supply—demand— price relationship was in balance. The veneer conversion industry was generally assured a "fair” profit or return on their investment. When the export market for logs develOped, the supply-demand-price equilibrium was disturbed. The length of time taken to restore equilibrium has been pro— longed by: (l) the temporary export limit on logs in 1964 (2) the inconsistent action of foreign buyers in the market place (3) the lack of innovative action by domestic veneer production and marketing units to meet this challenge (4) a somewhat erratic demand for domestic walnut veneer products since 1968. 120 e. Innovative Action. With system improvement our goal, the custom of marketing domestic veneer flitches intact should also be questioned. This practice has direct influence on marketing efficiency and raw material evalua— tion. The key to the general concept is in this fact Of processing: at each stage a single input unit is transformed into two or more output units. Each output unit may be a different product or different grade. The output units can be evaluated more precisely than can the input unit because they are one step nearer final use and can be inspected more completely. This concept suggests the study firm in— vestigate the alternative of processing individual flitches into products. Veneer should be allocated to that market which would maximize its dollar return. Certain flitches would be marketed intact; others divided into two or more products. This output information from an improved pro— cessing and marketing system could be used to improve the raw material evaluation and log—costing system. 2. Buying System Limitations a. Grading Half—logs. There are certain buying system improvements that warrant consideration if system improvement is our goal. Application of grading rules to half-logs should warrant consideration. Processing and 121 marketing is based on half-10g flitches. Why, then, limit evaluation to full logs? Evidence from this study indicates there are but two domestic veneer log quality classes. If half—logs were graded, four, rather than two, unit price levels could be established for each log. b. Buyer Training. Other system controls that warrant investigation are more thorough training of log buyers and develOping an incentive plan based on profit from logs purchased by the individual log buyers. LI TERATURE CI TED l) 2) 3) 4) 5) 6) 7) 8) 9) LITERATURE CITED Quigley, K. L. and R. D. Lindmark, A Look at Black Serv., Northeastern For. Walnut. U.S. For. Sta. Resource Bulletin NE—4, 28 pp., 1967. Exp. Hair, D. and A. H. Ulrich, The Demand and Price Situa— tion for Forest Products 1969—70. Division of Forest Economics and Marketing Research, Misc. Pub. No. 1165, 79 pp., 1970. Chapman, H. H. and W. H. Meyer, U.S. For. Forest Mensuration, McGraw Hill Book Co., Inc., New York, N.Y., pp., 1949. Forest Products Laboratory, 522 Overall Work Plan for Serv. Development of Logjand Bolt Grades for Hardwoods, U.S. For. Serv., For. Prod. Lab., Report TGUR-l6, 62 pp., 1958. Ostrander, M. D., A Guide to Hardwood Log Gradigq (Revised), U.S. For. Serv., NE. For. 50 pp., 1965. NeWport, Carl, C. R. Lockard and L. Vaughan, Log and Tree Grading as a Means of Measuring Quality, Exp. Sta., For. Serv. (Limited Distribution), 45 pp., 1958. U.S. Anon., Hardwood and Decorative Plywood, NBS Voluntory Product Standard PSSl—7l, pp., 1971. Stayton, C. L., R. M. Morden, Defects Indicators and Their Associated Interior Defects in Sugar Maple, Forest Products Journal, Vol 20, No. 2, pp. U.S. Dept. R. G. Buchman, of Com., 16 Exterior 55-58, Feb., 1970. Lane, P. H., Evaluating Log and Tree Quality for Wood Products, Forest Products Journal, Vol. pp. 89—93, March, 1963. 122 13, NO. 3, 10) 11) 12) 13) 14) 15) 16) 17) 18) 19) 20) 123 Gaines, E. M., Log and Tree Quality Concepts in the Computer Age, Society Of American Foresters, Pro- ceedings, 1965. , Analytical Methods for Studying Timber Quality From Tree to End Product, ProceedingilUFRO, Section 41, Oct., 1965. Henley, J. W., E. H. Bulgrin, H. H. Haskell and R. O. Woodfin, Jr., Work Plan For Hardwood Veneer Log and Bolt Grade Development, U.S. For. Serv., For. Prod. Lab. (Limited Distribution COpy), 13 pp., 1963. Bulgrin, Erwin H., Hardwood Tree Diagramming Manual, U.S. For. Serv., Central States For. EXp. Sta., Manual NO. 1, 48 pp., 1964. Moslemi, Ali A., Wood Color Evaluation: Some Tools and Considerations, School of Agriculture Publication No. 30, Southern Illinois University, Carbondale, 111., 19 pp., 1967. Draper, N. R. and H. Smith, Applied Regression Analysis, John Wiley & Sons, Inc., New York, N.Y., 1966. Freese, Frank, Linear Regression Methods for Forest Research, U.S. For. Serv., Research Paper FPL 17, Madison, Wis., 1964. Ezekial, M. and K. A. Fox, Methods of Correlation and Regression Analysis, Ed. 3, John Wiley & Sons, New York, N.Y., 1959. Anon., Least Squares-Multiple Regression, Ag. Exp. Sta., Library Prog. No. 7, Mich. State Univ. East Lansing, Mich., 128 pp., 1967. , Least Squares—Multiple Regression Variable Deletion, Ag. Exp. Sta., Library Prog. NO. 8, Mich. State Univ., East Lansing, Mich., 38 pp., 1967. , Least Squares—Multiple Regression Variable Addition, Ag. Exp. Sta., Library Prog. NO. 9, Mich. State Univ., East Lansing, Mich., 32 pp., 1967. 124 21) Sokal, Robert R. and F. James Rohf, Biometry: The Principles and Practice of Statistics in Biologiggl Research, W. H. Freeman and Company, San Francisco,' Calif., 776 pp., 1969. APPENDICES APPENDIX I Input variable number, description, measurement unit, code and format of program LOGAN. NO. DESCRIPTION UNITS CODE COL- FORMAT UMNS 1 Unique individual half— log number —-— IHLN 1—3 I3 2 Flitch number -—— IFLN 4—8 15 3 Disposition of other half-log Coded IDHLT 9 ll 4 Other half-log unique number —-— IQHL 10—12 I3 5 Log type (butt or upper) Coded LTP 13 ll 6 Stain level Coded ISTAIN 14 I1 7 Grain contrast level Coded ICON 15 ll 8 Color level Coded ICOLOR 16 ll 9 Annual ring pattern Coded IRPAT 17 11 10 Log freshness Coded IFRESH 18 I1 11 Log length Tenths ALLEN 19—21 F3.1 feet 12 Amount of log sweep Tenths ASWEEP 23—24 F2.1 inches 13 Direction of sweep Coded DSWEEPl 25 Fl.0 l4 Direction of sweep Coded DSWEEP2 26 Fl.0 125 APPENDIX I (cont'd.) 126 NO. DESCRIPTION UNITS CODE COL- FORMAT UMNS 15 Amount of crook Tenths ACROOK 27—28 F2.1 inches 16 Distance crook from Tenths ACROOK 29—31 F3.1 small end feet 17 Direction of crook Coded DCROOKl 32 Fl.0 18 Direction of crook Coded DCROOK2 33 F1.0 19 Spiral grain Tenths SPRGR 34—35 F2.1 inches 20 Pin knots visible #/S.F. PINK 36—37 F2.1 21 Diameter — Small End Tenths DIAS 38—40 F3.1 inches 22 Sap thickness — Small Tenths SAPTS 41—42 F2.1 End inches 23 Bark thickness — Tenths BARKTS 43—44 F2.1 Small End inches 24 Pith displacement, SE Tenths AHCS 45—46 F2.1 inches 25 Direction pith Coded DHCSl 47 F1.0 displacement, SE 26 Direction pith Coded DHCSZ 48 Fl.0 displacement, SE 27 Diameter — Large End Tenths DIAL 49—51 F3.1 inches 28 Sap thickness - Tenths SAPTL 52—53 F2.1 .Large End inches 127 APPENDIX I (cont'd.) NO. DESCRIPTION UNITS CODE COL— FORMAT UMN S 29 Bark thickness — Large Tenths BARKTL 54-55 F2.1 End inches 30 Pith displacement, LE Tenths AHCL 56-57 F2.1 inches 31 Direction pith Coded DHCLl 58 Fl.O displacement, LE 32 Direction pith Coded DHCL2 59 Fl.0 displacement, LE 33 Growth rate uniformity Coded IRING 60 F1 34 Visibility of new bark Coded BARKNV 61 Fl.0 growth 35 New bark color Coded BARKNC 62 F1.0 36 Old bark vigor Coded BARKOW 63 F1.0 (weathering) 37 Length, receiving Full RLENG 64—65 F2.0 foot 38 Diameter, receiving Full RDIA 66—67 F2.0 inch 39 Log grade, receiving Grade RGRADE 68 R1 40 Log grade, buyer Grade BGRADE 69 R1 41 Log buyers identifica— Coded IBUY 70-71 12 tion number 42 Buyer assigned lot —-— ILOT 72—74 I3 number 43 Geographical log source Coded IAREA 75-77 I3 128 APPENDIX I (cont'd.) NO. DESCRIPTION UNITS CODE COL- FORMAT UMNS 44 Flitch volume Actual FLFOOT 78-80 F3.0 Bd.Ft. 45 Bird Peck #/S.F. IBPECK 22 11 SECOND DATA CARD 46 No. 1 Repeated -—- IHLN 1—3 13 47 Log cost, actual $ & ¢ PRICEL 4—8 F5.2 dollars 48 Veneer length Tenths VLEN 9—11 F3.1 feet 49 Footage, veneer Sq.Ft. VFTAGE 12-15 F4.0 50 Veneer price ¢/S.F. VPRIC 16-17 F2.3 51 Veneer grade ——- VGRAD 18—19 R2 52 Number log defects -—— NUMD 20 11 53 Sector of defect ——- ISECT 21 Il location* 54 Beginning section of ——— IBEG 22-23 12 defect 55 Ending section Of —-- IEND 24-25 12 defect 56 Defect severity level ——— ILEV 26 12 times to accept up to ten defect specifications. *Defect information in columns 21 to 26 repeated nine APPENDIX II Variable classification and measurement unit of field data. Continuous: Variable Measurement Unit Length 0.1 ft. Sweep 0.1 in. Crook 0.1 in. Crook-distance from small end 0.1 ft. Diameter-small end 0.1 in. Sapwood thickness-small end 0.1 in. Bark thickness-small end 0.1 in. Pith displacement—small end 0.1 in. Diameter-large end 0.1 in. Sapwood thickness—large end 0.1 in. Pith displacement-large end 0.1 in. Discontinuous: Log defects number Ranked: Level Stain Grain contrast Color Annual ring pattern Log freshness Bird peck Bark growth visibility New bark color Old bark vigor WWWWWIPPUTUTUI 129 130 APPENDIX II (cont'd.) Attribute: Level Log type 2 Growth rate uniformity 2 Sweep direction sector Crook direction sector Pith displacement direction—small end sector Pith displacement direction—large end sector Defect location sector Defect location, beginning section Defect location, ending section Defect severity, level 2 Geographical log source 419 Computed: Measurement Unit Pin knots #/SF Spiral grain 0.1 inches in 2 ft. Veneer Yield Data: Length 0.1 ft. Area sq. ft. Unit price ¢/sq. ft. APPENDIX III Magnitude boundaries of ranked variables. Stain (Evaluated after end trim): = None = Evidence to 2% end area More than 2% to 10% = More than 10% to 50% = More than 50% thLUNJH ll Grain contrast level: Grain contrast is defined as the distinct shading of the annual rings that makes a grain pattern apparent. 1 = Uniform color of growth rings except it is several shades darker at the summerwood boundary. Desirable. 2 = (a) Only slight shading of summerwood boundary resulting in faintly apparent annual rings, or, (b) Dark summerwood boundary area resulting in pronounced annual ring appearance. 3 = (a) Lack of darkening of summerwood boundary resulting in uniform shade of annual rings. (b) Very dark late summerwood boundary re— sulting in extremely pronounced annual ring appear- ance. 4 = Distinct bands of several annual rings alternating light and dark shades with gradual shade transition. Little or no annual ring appearance. 5 = Distinct bands Of several annual rings of alternating light and dark shades and transition abrupt. Includes abrupt shade change between and along annual ring. 131 132 APPENDIX III (cont'd.) Color level: Masked by pathological stain. Grey to light brown. Light brown to medium brown. Medium brown to dark brown with slight and reds. = Range of colors, heavy to purple and reds. purpl blbtphahlo II Annual ring pattern: 1 Uniform — ring width, concentric pattern. 2 Slight — from distinguishable variation to 1/3 ring width variation. Slight variation from normal concentricity (to 1/4 inch per inch diameter). 3 = Medium - up to 100% variation of average ring width along and between annual rings. Medium variation from concentricity (1/4 to 3/4 inch per inch diameter). 4 = Wild — greater than medium limit in either or both characteristics. Log freshness: 1 No sign of stain or weathering. 2 Slight indication of presence of stain and weathering. 3 = Heavy presence of stain, weathering and checking of log end. Bird peck: O = None — no visible indication of Occurance. l = Slight — any indication of Occurance in an area not greater than 0.5 square foot. 2 = Heavy - any indication of Occurance in more than one location or one area greater than 0.5 square foot. Visibility of new bark growth: 1 = Not visible. 2 Slightly visible. 3 = Distinctly visible. I» 133 APPENDIX III (cont'd.) New bark color: 1 = Bright 2 = Medium 3 = Dull Old bark vigor: 1 = Fissures long, sh 2 = Fissures broken, 3 = Fissures short, b Geographical log source: 111 to 999, First digit id groups of counties within state dividual counties with group. information. allow, sharply edged deep, eroded edges ark cross broken entifies state, second, and third identifies in- Classified study corporation APPENDIX IV Output variable number, description, measurement unit, code and format of program LOGAN. NO. DESCRIPTION UNITS CODE COL- FORMAT UMNS 1 Unique individual half- ——— N 2—4 13 log number 2 Minimum length of --— CUT 5—6 F2.0 cutting allowed 3 Number section, cutting --— NS 7 11 width 4 L09 type Coded LTP 8 11 5 Stain level Coded ISTAIN 9 11 6 Grain contrast level Coded ICON 10 11 7 Color level Coded ICOLOR 11 11 8 Annual ring pattern Coded IRPAT 12 11 9 Log freshness Coded IFRESH 13 11 10 Bird peck Coded IBPECK 14 11 11 Growth ring uniformity Coded IRING 15 11 12 Log buyer identifica— Coded IBUY 16—17 12 tion 13 Geographical lOg Coded IAREA 18-20 13 source 14 Spiral grain Tenths SPRGR 21—24 F4.l inches 134 APPENDIX IV (cont'd.) 135 NO. DESCRIPTION UNITS CODE COL- FORMAT UMNS 15 Pin knots visible #/SF PINK 25—28 F4.1 16 Average bark vigor Coded AUBKFA 29-32 F4.1 rating 17 Amount of log sweep Tenths ASWEEP 33—36 F4.1 inches 18 Amount of log crook Tenths ACROOK 37-40 F4.1 inches 19 Mean sap thickness Tenths SAPTA 41—44 F4.1 inches 20 Mean pith displacement Tenths ANCA 45-48 F4.1 inches 21 Mean bark thickness Tenths BARKTA 49—52 F4.1 inches 22 Log length Tenths ALLEN 53-56 F4.1 feet 23 Diameter small end Tenths DIAS 57—60 F4.1 inches 24 Diameter large end Tenths DIAS 61—64 F4.1 inches 25 Receiving length Full RLEN 65—66 F2.0 foot 26 Receiving diameter Full RDIA 67—68 F2.0 inch 27 Receiving log grade -—- LT 69—70 I2 28 Log cost Dollars RLPRIC 71—76 F6.2 & cents APPENDIX IV (cont'd.) 136 NO. DESCRIPTION UNITS CODE COL— FORMAT UMNS SECOND CARD 29 Number 1 repeated ——— N 1—3 13 for identification 30 Number 2 repeated ——— CUT 4-5 F2.0 for identification 31 Number 3 repeated --— NS 6 11 for identification 32 Volume half—log Actual VOL 7-11 F5.1 Bd.Ft. 33 Corrected volume Actual CVOL 12—16 F5.1 Bd.Ft. 34 Volume clear wood, Actual CLOUTV 17—21 F5.1 surface division Bd.Ft. 35 Volume clear wood, Actual CLINV 22—26 F5.1 center division Bd.Ft. 36 Volume clear wood, Actual NETCUT 27—31 F5.1 both divisions Bd.Ft. 37 Per cent volume, Percent PCL 32—37 F6.1 3 both divisions 38 Total clear wood Actual TCW 38—43 F6.1 Bd.Ft. 39 Number of clear cuts -—— NCUT 44—45 12 I per half—log 40 Number of clear cuts ——— NUCUT 46—47 12 surface division 137 APPENDIX IV (cont'd.) NO. DESCRIPTION UNITS CODE COL- FORMAT UMNS 41 Number of clear cuts ——- NLCUT 48-49 12 center division 42 Length of minimum Tenths CMIN 50—53 F4.1 cutting encountered feet 43 Veneer footage Sq.Ft. VCTAGE 54—57 F4.0 recovered 44 Veneer price Cents VPRIC 58—63 F6.3 45 Veneer grade —-— IVT 64—65 12 46 Veneer length Tenths VLEN 66-69 F4.1 feet 47 Veneer value Dollars VVALU 70—74 F5.1 ¢ cents 48 Weighting factor for -—— WT23FAC 75-80 F6.1 sections APPENDIX V Variable classification and measurement unit after preliminary computations. Continuous: Variable Measurement Unit ft. in. Length Diameter—small end 0 1 O 1 Diameter—large end 0.1 in. Crook 0.1 in. Sap thickness—mean 0.1 in. Pith displacement—mean 0.1 in. Bark thickness—mean 0.1 in. Discontinuous: Log defects number Clear cuts—half log number Clear cuts-surface section number Clear cuts—center section number Ranked: Level Stain 5 Grain contrast 5 Color 5 Annual ring pattern 4 Log freshness 3 Bird peck 3 Growth rate uniformity 2 Bark vigor 3 Geographical log source 2 Log type 2 138 139 APPENDIX V (cont'd.) Computed: Pin knots Spiral grain NO./Sq. Ft. 0.1 in./ft. Sweep 0.1 in./ft. Volume-corrected B.F. Clear wood volume—surface section B.F. Clear wood volume-center section B.F. Clear wood volume-total B.F. Clear wood percentage Of total volume % Clear wood B.F. Clear cutting length—maximum 0.1 ft. Section weighting factor Ratio Veneer Yield Data: Length 0.1 ft. Area Sq. Ft. Unit price ¢/Sq. Ft. 140 APPENDIX VI Flow chart of computer program LOGAN. 141 C START D 0 READ E 9 DATA CONV RT (ONE LOG) ALPHANUMERIC LOG a VENEER GRADES TO SET NUMERIC VALUES DEFECT - ARRAY I I COMPUTE MEAN COMPUTE SAPWOOD a BARK LOG 8 SECTION THICKNESS, PITH VOLUME DISPLACEMENT I I COMPUTE BARK, SPIRAL COMPUTE DATA VALUE CORRECT PRINT 8 PUNCH LOG VOLUME OUTPUT FOR ABOVE DATA FACTORS ' I COMPUTE o ”0 CLEAR CUTTING LENGTH 8 VOLUME YES 0 C D I'IICH G IIIIIIIIIIIIIIII