Victor D. Phillips, Ph.D., Assistant Director; Audun E.
Tvedten, M.S., Graduate Student, College of Tropical Agriculture and Human
Resources, University of Hawaii at Manoa, Honolulu, HI 96822 U.S.A.
Wei Liu, M.S., Postgraduate Research Fellow, Oak Ridge National
Laboratory, Biofuels Feedstock Development Program, P.O. Box 2008, MS-6352, Oak
Ridge, TN 37831-6352 U.S.A.
Robert A. Merriam, M.S., Consulting Forester, 616 Pamaele Street,
Kailua, HI 96734 U.S.A.
From the Proceedings, Second Biomass Conference of the Americas: Energy, Environment, Agriculture, and Industry. Meeting held August 21-24, 1995, Portland, Oregon; published by National Renewable Energy Laboratory, Golden, Colorado.
Electricity produced from burning sugarcane bagasse has provided as much as 10 percent of Hawaii's electricity supply in the past. As sugarcane production has ceased on the islands of Oahu and Hawaii and diminished on Maui and Kauai, the role of biomass energy will be reduced unless economically viable alternatives can be identified. An empirical biomass yield and cost system model linked to a geographical information system has been developed at the University of Hawaii. This short-rotation forestry decision support system was used to estimate dedicated biomass feedstock supplies and delivered costs of tropical hardwoods for ethanol, methanol, and electricity production. Output from the system model was incorporated in a linear programming optimization model to identify the mix of tree plantation practices, wood processing technologies, and end-products that results in the highest economic return on investment under given market situations. An application of these decision-support tools is presented for hypothetical integrated forest product systems established at two former sugarcane plantations in Hawaii.
Results indicate that the optimal profit opportunity exists for the production of medium density fibreboard and plywood, with annual net return estimates of approximately $3.5 million at the Hamakua plantation on the island of Hawaii and $2.2 million at the Waialua plantation on Oahu. Sensitivity analyses of the effects of different milling capacities, end-product market prices, increased plantation areas, and forced saw milling were performed. Potential economic credits for carbon sequestration and wastewater effluent management were estimated. The favorable net return estimates and the carbon and wastewater credits suggest that commercial forestry ventures merit consideration for these sites in Hawaii. While biofuels are not identified as an economically viable component, energy co-products may help reduce market risk via product diversification in such forestry ventures.
Hawaii is undergoing profound land use changes due to urbanization and tourism, heightened environmental awareness and conservation of island biodiversity, and transition from traditional plantation agriculture to diversified agricultural industries that enhance agroecosystem sustainability. To help plan for a future that is consistent with local values and to implement effective land-use strategies during this period of rapid change, decision support tools are urgently needed. Our research team has evaluated the feasibility of short-rotation forestry on former sugarcane and pineapple plantation lands to manufacture a variety of wood products, including biofuels from wood chips.
To provide useful information to those interested in short-rotation forestry
as an alternative land use, our research team at the University of Hawaii has
developed a short-rotation forestry decision support system (SRFDSS) (
The SRFDSS has been used to estimate yield and delivered costs of wood
chips from hypothetical tree plantations on Kauai, Maui, Molokai, Hawaii, and
Oahu to specific bioconversion facilities on each island (
Figure 1
Figure 1 is a schematic overview of the integrated forest products components that are featured in the analysis. Because Eucalyptus saligna (Sydney blue gum) was determined to be more productive and cost effective than E. grandis (flooded gum) for both Waialua and Hamakua plantations, this species was chosen for subsequent system model analyses to develop 84 different combinations of growing space per tree, rotation age, thinning, and N-fertilizer for input to the LP model. Optimal spacing and rotation age for E. saligna chip production on Waialua and Hamakua plantations were identified using the SRFDSS model. This information was used to develop graphs of optimized management strategies and biomass supply curves for both plantations as managed for chip production only. Maps depicting the yields and delivered costs of E. saligna chips for both plantations were generated using HNRIS and the SRFDSS model.
A literature search was conducted for information on technical specifications, performance data, scale, and capital, fixed, and variable operating costs of a variety of wood processing technologies. Information on product prices, carbon sequestration by tropical hardwoods, and wastewater quality and quantity were obtained. Linear programming development and results of net return estimates of the optimized integrated forestry operations at the Hamakua and Waialua plantations are presented below.
Plantation management, wood processing technology, and end-product price variables were used to develop the LP model. Base and alternative cases were identified for the analysis (technical and economic assumptions in the LP model development as well as the results reported here are detailed in a manuscript being prepared for submittal to an appropriate peer-reviewed journal for publication).
The SRFDSS utilizes both intrinsic site conditions (elevation, temperature,
solar radiation, rainfall, and soil nitrogen and pH) and manageable factors
(growing space, rotation age, and N-fertilizer) to estimate mean tree diameter
at breast height (DBH) and stand productivity, as well as costs associated with
establishment, maintenance, thinning/harvesting, transport, and storage of wood
material. Because growing space (planting density) and rotation age variables
account for approximately 75 percent of the regression coefficient values for
E. saligna yield, with N-fertilizer application being the next most
important management variable, these three variables were employed in
developing the plantation management options for LP analysis. Options for zero,
one, two, and three thinnings were also incorporated into the LP model. A
feller-buncher and grapple-skidder provided the least-cost harvesting system as
described by Liu et al. (1992). Nitrogen fertilizer
application was derived from BDC recommendations (
The SRFDSS was used to calculate yields and costs for the different plantation management options at each plantation. Mean tree DBH estimates were calculated by the SRFDSS at the times of various thinnings and final harvest (rotation age). These DBH data were used to determine log distribution to suitable wood processing mills, e.g., chip and MDF milling, veneer milling, and saw milling. This log 'split' or percentage distribution of log types based on DBH was accomplished by selecting a certain growing space per tree (10 m2/tree) as being representative of the forest stand described in Table A.9.3 of Groome Poyry, Ltd. (1994).
For the plantation management options, the wood production yield and cost estimates developed using the SRFDSS are presented for (a) Hamakua plantation and (b) Waialua plantation. This information was used subsequently in the LP analysis to select the integrated forest products system (both plantation management option and wood processing technology and price option) that maximized net return on investment.
The wood processing technologies used in the LP analysis are those featured in the Hawaii Forestry Investment Memorandum (Groome Poyry, Ltd., 1994) at the same and different plant capacity scales: (a) saw mill for sawn lumber of various quality grades; (b) veneer mill (Meinan "Aristlathe" with drier and press) for plywood production; (c) medium density fibreboard (MDF) mill; and (d) chip mill for hardwood chip export. Through an extensive literature search and numerous contacts with national and international experts, reasonable data for each of these technology/price variables were developed and characterized by (1) capital requirement at various production capacities, (2) product recovery and loss described factors, (3) operating costs, (4) annual depreciation rate, and (5) average market price at the plant gate.
Prices for wood products were derived from Forest Products Prices, 1971-1990 (U.N. Food and Agricultural Organization, 1992) for plywood, MDF, and export chips. Sawn lumber prices were derived from Asian Timber as reported by Tisseverasinghe (1989). To calculate capital costs, an annual discount rate of 10 percent and an annual depreciation rate of 10 percent were used. Operating costs were calculated simply as a percentage of capital costs based on examples found in the literature.
Any of the wood processing/end-product price options used in the LP model contains a certain set of production processes, with a certain relationship between input and output for each process. The LP model chooses the optimal combination of these production processes including the optimal scale (production capacity) of operation. Base case #1 and base case #2 differ in processing plant capacity, with base case #2 having a chip mill, MDF mill, and veneer mill at one-half the capacity of base case #1 (the saw mill capacity remains the same). When the processing plant capacities are decreased, both capital and operating costs per m3 of end-product increase, i.e., economies of scale are evident, and are estimated from reports in the literature.
Alternative wood processing/end-product price options were used in LP sensitivity analyses to determine the effects of +/- 20 percent price changes (alternative case #1), plantation land area tripled and three out of four mill capacities increased (alternative case #2), and forced saw milling (alternative case #3) with a fixed annual production of 4,000 m3 sawn lumber in the integrated operation (all other assumptions remained the same as those in base case #1).
The utility and strength of the LP model are that a multitude of possible combinations of plantation management and production process options, which are subject to a set of resource limitations and production coefficients, can be evaluated accurately and expeditiously. The LP model finds the best possible combination for a given objective function, which may be either maximized or minimized. Because it is easy to change assumptions (coefficients and restrictions or bounds), the LP model facilitates the ability to conduct 'what-if' and sensitivity analyses. The LP software used for model calculations in this study was the LPS-867, Release 4.05 developed by Applied Automated Engineering Corporation.
The design criterion of the LP model was to maximize net return as the objective function. Net return was defined as the difference between the plant-gate revenue received from selling the products of the processing plants and the production costs of both the tree plantation and wood processing operations. The plantation and processing components were modeled as an integrated system, in which there are no internal prices (i.e., there is no income to the plantation operation and no cost for wood material to the processing operation). However, the LP model evaluates wood production costs in identifying the unique combination of a plantation management scheme and a set of processing plants that maximizes net return.
In the LP model runs, the plantation management options produce three different log types based on diameter size in various proportions. The logs may be utilized in one or more of four different types of wood processing plants. Within the scope of this project as the LP model exists presently, it describes a situation beyond the start-up period where both the tree plantation and the wood processing mills are operating at a steady state. This assumption means that the tree plantation is supplying a constant yield of logs of the same amount and log size distribution every year. This allows for maintaining a constant workforce, and for providing the processing plants with a stable, steady source of raw material. The LP model considers planting, thinning, and harvesting a certain number of hectares each year when selecting the optimal plantation management option. This number of hectares is equal to the total plantation area divided by the number of years from planting to final harvest (rotation age) for each particular option. The processing technologies are also modeled as a steady-state operation, as long as other variables, e.g., prices, are constant. Therefore, LP model calculates net return estimates only at a point in time after the start-up period is over. In addition to the revenue from the production of wood products, revenue may be obtained in the future from a "carbon sequestration credit" or from a "wastewater credit."
Results of the LP model calculations were developed for the base and alternative cases as annual end-product production (1000 m3/yr), net return (1000$), and carbon and wastewater credits (1000$) for both plantations. Also, the plantation management option and the wood processing technology and price option selected, area harvested (ha), harvested yield (1000 m3), net return per end-product unit ($/m3), net return increase per unit required for the processing technology to become competitive ($/m3), shadow price of land ($/ha), and carbon sequestered in end-products (Mg C) are featured. By applying various restrictions either on the resources (e.g., land area available) or directly on the amount of end-product produced by each wood processing mill, sensitivity analyses were performed. In the two base cases, mill capacity was varied to determine the effect of plant processing scale on net return. In the alternative cases, the effects of price change, expanded plantation size and increased plant processing capacity, and forced saw milling were featured. Table 1 summarizes the annual net return results for all cases. Discussion of the results are presented by plantation below.
| Linear Programming Option | Hamakua Plantation, Hawaii (Annual net return) |
Waialua Plantation, Oahu (Annual net return) |
|---|---|---|
| Base Case #1 | $3,490,000 | $2,211,000 |
| Base Case #2 | $1,085,000 | $836,000 |
| Alternative Case #1a | $8,615,000 | $5,353,000 |
| Alternative Case #1b | [none; negative solution] | $12,000 |
| Alternative Case #2 | $23,390,000 | $14,522,000 |
| Alternative Case #3 | $3,307,000 | $2,028,000 |
To determine "how far" a certain non-selected production process is from being selected as part of the optimal solution, the LP model calculates the quantitative amount by which the contribution of the non-selected production to the objective function must change to become selected. This amount is expressed as the positive price change in dollars per end-product unit that is required for one of the non-selected productions to become selected as part of the optimal solution. This LP output is reported as net return increase per end-product unit required for the processing technology to become competitive. For example, in base case #1, for export wood chips to become competitive the net return would have to be $82/m3 ($28/m3+ $54/m3), and the net return for sawn lumber would have to be $110/m3 ($64/m3 + $46/m3).
Another interesting output feature from the LP model is the shadow price of land, which is the increase in annual net return if one more unit of land is added ($/ha). For example, in base case #1, each additional hectare of land would generate $449/ha at Hamakua and $326/ha at Waialua. Considering the final harvest times, the total increase in net return would be $8,980 ($449 X 20 yrs) over the 20-year rotation at Hamakua and $4,890 ($326 X 5 yrs) over the 15-year rotation at Waialua.
For base case #1, the annual net return is $3,490,000, where 35,000 m3/yr of MDF and 22,000 m3/yr of plywood are produced. Base case #2 features a 50 percent reduction in mill capacities with the same quantities of MDF and plywood produced as in base case #1 (in either case wood supply is the limiting factor, not mill capacity). The results of base case #2 indicate that the annual net return is approximately $1 million. This decrease in annual net return is attributed to diseconomies of scale associated with the smaller processing plants. For alternative case #1, a +20 percent increase in end-product prices increases the annual net return to $8,615,000, and a -20 percent decrease in prices results in a negative annual net return. By tripling the land area to 23,334 ha and utilizing processing mills with larger capacities, the LP model reported the annual net return to be 23,390,000. While economies of scale would be expected to improve plant processing return on capital, this is not necessarily the situation for the expanded plantation operation because of longer hauling distances and greater variability in site productivity. For alternative case #3, in which the LP model was forced to produce 4,000 m3/yr sawn lumber, the annual net return is $3,307,000, or $183,000/yr less than without forcing the saw mill. The disadvantage of "losing" the $183,000/yr may be countered by "gaining" product diversification to hedge against market risks.
With 45 percent of stem dry weight stored as carbon (
For base case #1, the annual net return is $2,211,000, where 21,000 m3/yr of MDF and 14,000 m3/yr of plywood are produced. With the reduced mill capacities (but the same quantities of MDF and plywood produced as in base case #1), base case #2 results indicate that the annual net return is $836,000. As for Hamakua, this decrease in annual net return is due to scale impacts of the smaller processing plants. For alternative case #1, a +20 percent increase in end-product prices increases the annual net return to $5,353,000, and a -20 percent decrease in prices decreases the annual net return to $12,000. By tripling the land area to 20,370 ha and utilizing processing mills with larger capacities, the annual net return increases to $14,522,000. However, for the same reasons as explained for Hamakua, this amount is artificially inflated because longer hauling distances and greater variability in site productivity are not accounted. For alternative case #3, in which the LP model was forced to produce 4,000 m3/yr sawn lumber, the annual net return is $2,028,000, or $183,000/yr less than without forcing the saw mill. Again as with Hamakua, the addition of the saw mill to increase end-product diversification may help hedge against market risks.
With 45 percent of stem dry weight stored as carbon (
The utility of the SRFDSS developed previously (to estimate yield and delivered cost of. tropical hardwood from parcels of land identified as both suitable and available for species-specific tree plantations in Hawaii) has been extended via linear programming capability (to estimate net returns of optimized combinations of wood products from specific plantations). The analytical results provide decision-makers who are contemplating alternative land uses on Hamakua and Waialua plantations readily useful information for scoping forest product investment strategies. The net return estimates for both plantations suggest that commercial forestry ventures merit consideration at these sites in Hawaii. Energy co-product(s) as part of potential forestry ventures will be determined primarily on their demonstrated contribution to net return on investment and secondarily to reduced market risk via product diversification.
This research was supported by the U.S. Department of Energy and administered by the National Renewable Energy Laboratory through subcontracts to the Hawaii Natural Energy Institute, the U.S. Department of Agriculture McIntire-Stennis Cooperative Forestry Research Program, and the State of Hawaii Division of Forestry and Wildlife.
Evans, J. 1982. Plantation Forestry in the Tropics. Oxford University Press. Oxford, London.
Groome Poyry, Ltd. 1994. Forestry Investment Memorandum, with Appendices. Contract to State of Hawaii Dept. Business, Economic Development and Tourism.
Liang, T., Khan, A., Phillips, V. D., Liu, W. 1994. Hawaii natural resource information system: a tool for biomass production management. Biomass and Bioenergy 6(6): 431-441.
Liu, W., Phillips, V.D., Singh, D. 1992. A spatial model for the economic evaluation of biomass production systems. Biomass and Bioenergy 3(5): 345-356.
Liu, W., Merriam, R.A., Phillips, V.D., Singh, D. 1993. Estimating short-rotation Eucalyptus saligna production in Hawaii. Bioresource Technology 45: 167-176.
Merriam, R.A., Phillips, V.D., Liu, W. 1995. Early diameter growth of trees in planted forest stands. Forest Ecology and Management (in press).
Phillips, V.D., Liu, W., Merriam, R.A., Bain, R.L. 1995. Short-rotation forestry as an alternative land use in Hawaii. Biomass and Bioenergy (in press).
Phillips, V.D., Liu, W., Merriam, R.A., Singh, D. 1994a. Biomass system model estimates of short-rotation hardwood production in Hawaii. Biomass and Bioenergy 5(6): 421-429
Phillips, V.D., Liu, W., Merriam, R.A., Singh, D. 1994b. Potential for short-rotation intensive-culture hardwood production in Hawaii, Agricultural Systems 46: 33-57.
Phillips, V.D., Merriam, R.A., Singh, D., Khan, M.A. 1993. Land available for biomass crop production in Hawaii. Agricultural Systems 43: 1-17.
Phillips, V.D., Tvedten, A.E., Liu, W., Merriam, R.A. 1995. Forestry investment opportunities in Hawaii. In Proceedings of the Hawaii Agriculture: Positioning for Growth Conference, Honolulu, Hawaii, April 5-6, 1995.
Singh, D., Phillips, V.D., Merrian R.A., Khan, M.A., Takahashi, PK. 1993. Identifying land potentially available for biomass plantations in Hawaii. Agricultural Systems 41:1-22.
Tisseverasinghe, K. 1989. The question of processing of logs and added value. Klinkii 4(l): 55-63.
United Nations Food and Agriculture Organization. 1992. Forest Products Prices 1971-1990. FAO Forestry Paper 104. FAO. Rome, Italy.
Whitesell, C.D., DeBell, D.S., Schubert, T.H. 1992. Short-Rotation Management of Eucalyptus: Guidelines for Plantations in Hawaii. Gen. Tech. Rep. PSW-GTR-137. Albany, CA: Pac. SW Res. Sta. U.S. Department of Agriculture Forest Service. 30 p.