CC BY
20
O 00
o
VJ5
O ■<t
O rí
mound pit scari untreated Figure 7. Cover degree of vegetation
6. Conclusion
In this study, the result shows the order of survival rate among planted seedlings from high to low was pit > scarification = untreated > mound. This result would be caused by soil moisture content because of quite dry condition on this test site. Though this result can be referred to silviculture in dry condition, it is expected that different result will be found in humid or mesic condition. Thus it is necessary to do similar study on the hillside or the bottom of slope, or using different species.
References
1. Stjernberg, E. I. (1988) A study of manual tree planting operations in central and eastern Canada. 42pp, FERIC Technical Report 79, FERIC, Pointe-Claire, Canada.
2. Stjernberg, E. I. (1991) Planter productivity in prepared and unprepared ground: A case study. 6pp, FERIC Technical Note 162, FERIC, Pointe-Claire, Canada.
3. Yamada, T. And Endo, T. (2003) Development of the automatic planter for containerized seedlings, Proceedings of the 54th meeting of Kanto brunch Japanese Forest Society: 239-240 (written in Japanese).
4. Stjernberg, E. I. (1985) Tree planting machines -A review of the intermittent - furrow and spot planting types-. 118pp, FERIC, Pointe-Claire, Canada.
5. Mashita, I. (1957) Soil property - Forest soil and its measurement-:102-120, Rinya Ko-saikai, Tokyo, Japan (written in Japanese).
6. Gonna, M. A. V. D. (1992) Excavator attachments for site preparation in British Columbia. 8pp, FERIC Technical Note 180, FERIC, Pointe-Claire, Canada.
7. Seamus, P. S. And Parker, P. (1992) Development of a silvicultural mounding attachment. 6pp, FERIC Technical Note 183, FERIC, Pointe-Claire, Canada.
8. Boeken, B. And Shachak, M. (1994) Desert plant communities in human-maid patches-implications for management. Ecological Applications 4-4:702-716.
Researcher Bruce TALBOT - The Danish Centre for Forest,
Landscape and Planning
THE INFLUENCE OF FOREST STAND SIZE AND LOCALITY ON THE OPERATIONAL EFFICIENCY OF MECHANISED FOREST
OPERATIONS
The sizes of stands on which forest operations are carried out upon, and the distances between pairs of stands, were analysed for four regions comprising all Danish state forests.
УкраТнський державний лкотехшчний унiверситет
Methods of assessing network density included single-linkage cluster analysis, a shortest path density estimate, as well as a simulation of relocation for a harvester and forwarder. Only marginal differences in machine utilisation were found.
Наук. ствроб. Брюс ТАЛБОТ - Датський центр лку,
ландшафту i планування, Датя
Вплив таксацшних параметр1в та розмщення лкового масиву на
1 • •• • • •
ефектившсть машиннот загот1вл1 л1су
Таксацшш характеристики лiсоексплуатацiйних масивiв та вщсташ мiж ними проаналiзовано в чотирьох регюнах, що охоплюють Bci Датсью державнi лiси. Мето-ди ощнювання густини люотранспортно'1' мереж включали кластерний аналiз, визна-чення найкоротшо'1' вщсташ, а також моделювання руху люозаготсвельно'1' i люотран-спортно'1' машин. Було знайдено тшьки граничнi вiдмiнностi у використанш лiсових машин.
Introduction
The level of spatial dispersion of raw materials distinguishes the production environment in forest operations from most industrial production settings. The necessity of having to frequently relocate production units between forest stands incurs a direct transport cost, a set-up cost, an efficiency reduction during the familiarisation phase in a new stand, as well as a subsequent indirect cost of reduced machine utilisation levels.
To some degree, forest operations managers have varying potential for obtaining high utilisation levels on their machines, owing simply to the differences in the underlying structure, or topology, of the forest. Distances between stands, and the size of destination stands play an important role in determining how much time is spent in relocating machinery, as against producing timber (Talbot et al. (2003).
Spatial constraints on forest harvesting have been addressed from a number of perspectives in the References. Both aesthetic and biodiversity issues have been solved for by clustering harvesting through periodic blocks, using algorithms which minimise 'interior edges' (Gustafson (1996) Gustafson (1998).). Economically optimal harvest scheduling can also handle spatial data, for example in the form of an internalised numerical adjacency measure (Tarp and Helles (1997)). However, the majority of procurement studies utilising GIS and explicit spatial analysis deal with biomass supply e.g. Bjorheden (1997); Borjesson and Gustavsson (1996a); Borjes-son and Gustavsson (1996b); Downing and0 Graham (1996); Forsberg (2000); Gustavsson and Borjesson (1996); Noon and Daly (1996).
The aim of this study is to quantify the effect that the underlying forest stand topology has on the potential to achieve reasonable levels of machine utilisation. Four machine-rings, owned by the Danish Forest and Nature Agency, were chosen as the administrative units of analysis. These machine-rings service all state forests in the country, and are located in geographic regions referred to here as North, Mid, South (all on the Jutlandic peninsula) and East, which covers the island of Zealand (Figure 1). It is therefore, a strictly comparative study.
Figure 1. The four operational areas covered by Figure 2. A map of region 'East' state-owned machine rings in Denmark. The (1:1 200 000), the black quadrants
blackened polygons represent coniferous stands represent the 1 km2 grid cells that
included in the study include at least one coniferous stand
Materials and methods
Universal Transverse Mercator (UTM zone 32 ED 50) co-ordinates of the centroids of all coniferous stands within each of the four operational areas, (Figure 2) were determined. Coniferous stands were considered to represent the forest area of interest, as fully mechanised cut-to-length (CTL) operations, are not yet commonplace in hardwood stands. In the forest management database used, individual stands were not directly linked to the forest road network, or to the national road database. Therefore, to overcome the problem of calculating distances between stands, a grid of 1 km cells was draped over the entire country, using a GIS. All coniferous stands were then accrued to the mid-point of a grid cell. The mid-point of each cell was linked through a line coverage to the national road database. Because of the very high public road density in the country (1-1.5 km km ), misrepresentation associated with this approach, on a regional level, was considered minimal.
The active grid cells were read into the network analyst programme in ArcView®. Using the national road database, a script was written to generate matrices of road distances between each pair of active grid cells. Together with data on forest densities, table 1 shows the number (n), of active grid cells in each region. The number of unique shortest routes between all pairs of grid cells is equal to half the matrix of n -n.
yKpaiHCbKHÖ .icp^aBUMÜ .ricoTexmHHMH ymBepcMTeT
Table 1. Overview of the extent of the operational areas, forest densities,
and grid statistics within each of the four regions
Region 2 Area (km ) Total forest (ha) Coniferous forest (ha) Density (ha.km-2) No. of grid cells
North 11 559 50 290 23 897 2.07 722
Mid 8 491 40 395 23 488 2.77 639
South 13 093 41 635 13 731 1.05 663
East 9 251 37 606 8 611 0.93 576
Owing to the high road density mentioned above, and the consequently large number of nodes in the road network, matrix generation was taxing, requiring 12 to 20 hours per region, utilising a 1.8 GHz processor. This prohibited stand level computations, which in the case of region North, would require the calculation of over 40 million shortest routes.
Cluster analysis
Hierarchical cluster analysis was carried out on the distance matrices using the cluster procedure in SAS® (Anon. (1989)). This procedure is commonly used in similarity analyses using co-ordinate data, potentially avoiding having to calculate the distance tables. However, because of large variation in the difference between euclidean and road distances amongst, and within regions (Figure 3), the analysis was done on road distance data. The single-linkage method of clustering was chosen as it is most relevant to the purpose of the study, i.e., classifying the effect of distance between stands themselves, and not, for example, between cluster centro-ids. The single-linkage method can be defined by:
DKL = minieCK min jeCL d(xi, xj ) (1)
which states that the distance between any two clusters (K, L) is given by the shortest distance of all possible combinations of elements in those clusters.
In a second analysis, the distance matrix was weighted by the inverse of the area at the destination. Thus, grid cells with small concentrations of coniferous forest area are penalised in a manner reminiscent of the expense of relocating forest machines to small stands, or large but distant stands.
Simulation
A Monte-Carlo style simulation was carried out in SAS. This involved generating stands, harvesting and forwarding timber, then relocating the machines to the next stand. Time consumption was calculated for each activity. The process was repeated 2500 times for each region. Each element of the simulation is described separately below.
Stand Generator
Stand sizes (ha) were generated from Weibull distributions fitted to the empirical GIS data, the parameters of which are shown in Table 2.
Mean extraction distances were calculated as the square root of the size of the stand generated by the Weibull function (Aedo-Ortiz et al., 2000), with a standard deviation arbitrarily set at 20 % of the mean. Individual tree volumes and the number of stems removed per hectare differ for each of three thinning types, as in
Talbot et al. (2003), originally derived from (Anon., 2001). In first thinnings they
3 _1 _1
are normally distributed around 0.05 m stem" 1 (1050 stems ha"1) incrementing
3 _1 _1 3 _1 _1
through 0.16 m stem (560 stems ha ) to 0.27 m stem (390 stems ha ) in the third thinning. Thinning number is determined from a uniform distribution.
Table 2. Parameters of empirical stand size distributions, by region
Parameter North Mid South East
Number of stands (n) 9172 7698 7231 5068
Mean size (ha) 2.61 3.05 1.89 1.70
Shape parameter: Weibull. 2.592 2.962 1.912 1.766
Scale parameter: Weibull 0.990 0.943 1.015 1.0947
Kolmogorov-Smirnov' 0.082 0.078 0.079 0.070
*at 99 % certainty.
Time consumption model
A time consumption regression model for a harvesting system comprising a harvester and forwarder were derived by Talbot et al. (2003) and implemented directly in the simulation as follows:
Y=p 0+^1 X + 2 X 2+p 3 X 3, (2)
_3 3 _1
where: Y _ System time-consumption (min m ); X1 _ Harvest volume (m ha ); X2 _ Stem-volume (m ); X3 = Lead distance (m). Parameters are given in Table 3.
Table 3. Parameters for the harvesting system time consumption model
Thinning Pc Pi P2
1st 25.72 _0.07023 _119.3
2nd 12.95 _0.02026 _16.34
3rd 9.820 _0.01507 _4.137
Graphical representations of the influence of each variable are given in figures 3 and 4 (Talbot (2002)). The time consumption models are used in determining utilisation rates on the machines. Equation (2) essentially provides E15 hours, which are effective hours including delays of up to 15 minutes (Samset et al. (1978)). In this analysis, E15 hours are equated to Productive Work Time (PW), as proposed by Bjorheden et al. (1995), although this is more representative of E0. However, doing this allows for a comparison at Work Time (WT) level which comprises both PW and Supportive Work Time (SW), here only measured as relocation. Thus utilisation is measured as the ratio of PW to WT.
Relocation
Because the distributions of distances between grid cells are multimodal, and therefore not easily fitted to a common distribution, cumulative density functions (Figure 3) were used, applying the inverse transformation method described by Hillier and Lieberman (2001). This involves generating a random, uniformly distributed number which represents the probability of an observation, then reading the value from the x-axis. The density functions were based on 1 km interval histograms, plotted from the original distance matrices.
yKpaÏHCbKHH icp^aBMMM .ricoTexMiHMMH yHÏBepcHTeT
90
"me (m3/ha)
(m3/ha)
Figure 3. Harvester productivity as a Figure 4. Forwarder productivity as a
function of harvest volume, and stem volume function of harvest volume and lead distance
Relocation refers to the movement of machines from one working tract (ob_3
ject) to another, and allows for specific time consumption (min.m ) to be calculated. This highlights the negative effects of long distances or small volumes (object volumes) on machine utilisation rates. The relocation model is duplicated directly from Talbot et al. (2003)and illustrated in Figure 4. At distances under 20 km, relocation always occurs under own power, while between 20 and 40 km, 50 % of the relocations are done by low-bed truck., which also accounts for all relocations exceeding 40 km. Movement on road occurs at an average velocity of 15 km hr_ !(under own power) alternatively 60 km.hr_1 (low-bed truck), and each move incurs a fixed time penalty of 30 minutes per machine. The'set-up' time is constant irrespective of relocation distance, and includes preparing to relocate as well as orientation on arrival in the new stand. Relocating outside of normal working hours does not influence machine utilisation. A random timestamp ranging between 10:00 and 19:00 was used to generate task-completion time. The rationale being that tasks that could have been completed before 10:00 would have been done so on the evening prior to the move (16:00-19:00 agreed overtime), and therefore only the time period 10:00 to 19:00 is of interest. A variable was used to retain the frequency and distance of low-bed truck relocations.
100
kx 80
,Q O
O
100 125 150 175 200 225 250
Road Distance (km)
Figure 5. Cumulative density functions for relocation distances for each region
Figure 6. Flow diagram of the relocation model
Shortest Path
Network Analyst in ArcView was used in determining the shortest path linking all stands within each region.
Results
For the non-weighted analysis, region 'East' has the lowest mean distance between stands (Table 4). This can be corroborated in Figure 7, where the majority of stands are 'joined' under the 10 km mark. The area-weighting procedure reverses the ranking, making 'North' the most dense region, while 'Mid' which has the largest stands, is less affected.
The shortest route density estimate produced surprisingly similar results in terms of distance, but differences in stand sizes mean that in region East, one has to travel 0.24 times further per hectare of coniferous forest.
Despite differences in stand sizes and relocations distances amongst regions, there were no discernable differences in machine utilisation levels. In thinnings, forwarders have a considerably higher productivity than harvesters.
yKpaiHCbKHH icp^aBMMM .ricoTexMiHMMH yMiBepcMTeT
100 90 — 80 — 70 60 50 — 40 30
20 —I
100 90 — 80 — 70 — 60 — 50 40 30 20 10 0 —1
L.............................HUM...............iJlllift^lgMliiiMM^^MllBlteiJ^ 6 ill
100 90 80 70 60 50 40 30 20 10 0
siliillirainllii
100 90 80 70 — 60 — 50 — 40 — 30
ImmillortlllrHliliLiy'llilllillllliffl
»31 jjiifiEffil
Figure 7. A graphic representation of the results of the non-weighted cluster analysis
HiivKOBiiii BiciiiiK, 2004, BHn. 14.3
Table 4. Results of the cluster analysis and shortest route density estimate
Results North Mid South East
Cluster Analysis
Mean distance between obs non weighted. 90.2 52.5 84.9 49.4
Mean distance between obs weighted 17.7 19.2 39.4 38.7
Shortest Route
Distance (km) 2487 2191 2684 2083
Density (km ha-1) 0.10 0.09 0.19 0.24
Table 5. Time consumption for harvesting, forwarding, and relocation, by stand, as well as utilisation levels for the harvester and forwarder
Region Harvester Hrs. Forwarder Hrs. Reloc hrs. Utilisation harvester Utilisation forwarder
North 7.36 4.18 1.78 0.80 0.70
Mid 7.13 4.03 1.97 0.78 0.67
South 7.68 4.28 1.75 0.81 0.71
East 7.97 4.50 1.99 0.80 0.69
Discussion
The cluster analysis chosen here provides a good graphical overview of operational areas, effectively showing both density and distance between stands. The single-linkage method is too simple to use in further analysis, for example, for generating solution sets for a sequencing algorithm.
The simulation failed to show any significant differences in the effect forest topology might have on machine utilisation. This could partly be explained by the relatively high set-up time at each stand (30 min.) which possibly disguises some variation.
References
1. Aedo-Ortiz, M, Olsen, E D, and Kellogg, L D (2000). Simulating a harvester-forwarder softwood thinning: a software evaluation. Forest Products Journal 47(5) 36-41.
2. Anon. (1989). SAS/STAT User's Guide Version 6. 4th ed. SAS Institute, Cary, NC. 846p.
3. Anon. (2001). Gallringsmallar, Sodra Sverige [Thinning Prescriptions, Southern Sweden]. ed. The Swedish National Board of Forestry. 35p.
4. Bjorheden, R (1997). Studies of Large Scale Forest Fuel Supply Systems. Swedish University of Agricultural Sciences
5. Bjorheden, R, Thompson, M, and Rickards, J (1995). Forest Work Study Nomenclature, WP 3.04.02. 1-16.
6. Borjesson, P. and Gustavsson, L. (1996a). Biomass transportation. Renewable Energy 9(1-4) 1033-1036.
7. Borjesson, P. and Gustavsson, L. (1996b). Regional production and utilization of biomass in Sweden. Energy 21(9) 747-764.
8. Downing, M. and Graham, R. L. (1996). The potential supply and cost of biomass from energy crops in the Tennessee valley authority region. Biomass & Bioenergy 11 (4) 283-303.
9. Forsberg, G. (2000). Biomass energy transport - Analysis of bioenergy transport chains using life cycle inventory method. Biomass & Bioenergy 19(1) 17-30.
10. Gustafson, E. J. (1996). Expanding the scale of forest management: Allocating timber harvests in time and space. Forest Ecology and Management 87(1-3) 27-39.
11. Gustafson, E. J. (1998). Clustering timber harvests and the effect of dynamic forest management policy on forest fragmentation. Ecosystems 1(5) 484-492.
12. Gustavsson, L. and Borjesson, P. (1996). Biomass utilisation and transportation demands. Renewable Energy 9(1-4) 1037-1040.
УкраТнський державний лкотехшчний унiверситет
13. Hillier, F S and Lieberman, G J (2001). Introduction to Operations Research. 7th ed. McGraw-Hill, New York. 1214p.
14. Noon, C. E. and Daly, M. J. (1996). GIS-based Biomass Resource Assessment with BRAVO. Biomass & Bioenergy 10(2-3) 101-109.
15. Samset, I, Clausen J, Mikkonen E, and Andersson S (1978). Forest Work Study Nomenclature. 1-129.
16. Talbot, B (2002). The Influence of Forest Stand Topology on the Operational Efficiency of Mechanised CTL operations. In Yoshimura (ed.), Proceedings of the International Seminar on New Roles of Plantation Forestry Requiring Appropriate Tending and Harvesting Operations. 242-248.
17. Talbot, B, Nordfjell, T., and Suadicani, K (2003). Assessing the Utility of Two Integrated Harvester-Forwarder Machine Concepts through Stand-Level Simulation. International Journal of Forest Engineering 14(2) 31-44.
18. Tarp, P. and Helles, F. (1997). Spatial optimization by simulated annealing and linear programming. Scandinavian Journal of Forest Research 12(4) 390-402.
19. Ohman, K. and Lamas, T. (2003). Clustering of harvest activities in multi-objective long-term forest planning. Forest Ecology and Management 176(1-3) 161-171.
Dr. int. Pawel TYLEK1; prof. dr. hab. Jozef WALCZYK - Agriculture
University in Cracow
MECHANICAL SEPARATION OF SEEDS
Economic aspects of nursery production require seeds of high genetics quality and high germination ability. Thanks to the separation process, full and well developed seeds are separated from empty and damaged ones. Nevertheless, this process cannot cause loss of lighter seeds (the presently used separation methods often lead to such results), as owing to the genetics variability, big and heavy seeds are as important as small and light ones. Aiming at the optimisation of the seeds mechanical separation process, from among many distinguishing features such a feature should be chosen that for the given seed lot, secures the shortest technological line for carrying out cleaning and grading The paper presents designs of two separators for forest tree seeds that were made in Department of Forest Work Mechanisation of Agricultural University of Cracow: pneumatic separator with a vertical air channel and column for separation of seeds, based on the difference in their elasticity.
Keywords: forest tree seeds, separation properties, sorting process
Др. Павел ТИЛЕК; проф. Др. ЙозефВАЛЬЧИК-Аграрний ун-т Кракова, Польща
Мехашчне вщдшення насшня
EKOHOMi4Hi аспекти люовщновлення вимагають насшня високо'1 генетично'1 якосп i добро'1 здатносп проростання. Завдяки процесу вщдшення, повш i добре роз-винут зерна вщокремлюються вщ порожшх i пошкоджених. Проте, цей процес мо-же викликати втрату бшьш легких зерен (методи вщдшення, що тепер використову-ються, часто приводять до таких результат), як наслщок, до мшливосп генетичних властивостей. Велик i важю зерна таю ж важлив^ як i малi та легю. Напрям оптимь зацп мехашчного процесу вщдшення зерен повинен бути вибраний так, щоб забезпе-чити найшвидше i найяюсшше очищення та сортування насшня. В статп представлено два роздшьники насшня люових дерев, як розроблеш на кафедрi мехашзацп ль сових роб^ Аграрного ун-ту Кракова: пневматичний роздшьник з вертикальним по-в^ряним каналом i колона для вщдшення зерен, принцип дп яких базусться на рiзни-щ еластичносп насшин.
Ключов1 слова: насшня люових дерев, властивосп роздшення, процес сортування
1 Department of Forest Works Mechanisation. E-mail: rltylek@cyf-kr.edu.pl. tel.: +4812 662 50 23. Al. 29-Listo-pada 46. 31-425 Krakow
