The Effect of Location and Facility Demand on the Marginal Cost of Delivered Wood Chips from Energy Crops:
A Case Study of the State of Tennessee

From the Proceedings, Second Biomass Conference of the Americas: Energy, Environment, Agriculture, and Industry; pages 1324-1333. Meeting held August 21-24, 1995, Portland, Oregon; published by National Renewable Energy Laboratory, Golden, Colorado.

Cost-supply curves for delivered wood chips from short rotation woody crops were calculated for 21 regularly-spaced locations spanning the state of Tennessee. These curves were used to systematically evaluate the combined effects of location and facility demand on wood chip feedstock costs in Tennessee. The cost-supply curves were developed using BRAVO, a GIS-based decision support system which calculates marginal cost of delivering wood chips to a specific location given road network maps and maps of farmgate prices and supplies of woody chips from short rotation energy crops.

Marginal costs of delivered chips varied by both facility location in the state and facility demand. Marginal costs were lowest in central Tennessee unless the facility demand was greater than 2.7 million dry Mg per year (3 million dry tons per year) in which case west Tennessee was the lowest cost region. Marginal costs rose rapidly with increasing facility demand in the mountainous eastern portion of the state. Transportation costs accounted for 18 to 29% of the delivered cost and ranged between $8 and $18/dry Mg ($7 and $16/dry ton). Reducing the expected farmer participation rate from 100% to 50% or 25% dramatically raised the marginal costs of feedstock supply in the east and central regions of the state. The analysis demonstrates the need to use geographically-specific information when projecting the potential costs and supplies of biomass feedstock.

The purpose of this study is to systematically quantify both regional differences in the cost of supplying wood chips to biomass energy facilities in the state of Tennessee and the effect of facility demand on those differences. While it is well known that biomass feedstock cost-supply curves will be location-specific, no study has systematically quantified the expected geographic variation in feedstock supply curves for a region. Without some understanding of the geographic variation, it is impossible to assess the magnitude of error of generalizing from one location to another or using aggregate data to make site-specific estimates. This study goes the next step and demonstrates a methodology for quantifying the geographic variation in feedstock supply curves using the state of Tennessee as a case study.

The study extends previous studies by Noon, Graham and Downing in this region (Graham and Downing 1995, Downing and Graham 1995, Graham and Downing 1993, Downing and Graham 1993, Noon et.al. 1993). These studies mapped projected farmgate prices and woody biomass supplies(2) at a county-level using a breakeven analysis which took into account the density of crop and pasture land, expected yields of woody crops, and current profitability of crop and pasture land.(3) In conjunction with the biomass production estimates, a GIS-based decision support system called BRAVO was developed and applied to assess the probable cost of supplying wood chips from SRWC plantations to specific Tennessee Valley Agency (TVA) power plants in this region. BRAVO uses road networks and farmgate price and supply maps in conjunction with a transportation model to develop delivered cost supply curves for a given location or collection point.

In this study BRAVO is used to develop delivered cost-supply curves for 21 locations evenly spaced across the State. These supply curves are then systematically examined at 10 facility demand levels corresponding to a range of likely power plant or ethanol plant sizes. Regional differences in costs at each demand level are quantified as are (1) portion of the cost attributable to transporting biomass from the farm to the facility, and (2) effect of reducing farmer participation rate (the proportion of the land that is converted to SRWC plantations if the projected farmgate price for wood from that land is met).

Selection of Sites for Supply Curves

A grid of 21 points evenly spaced across the state was used to define the locations for the BRAVO runs (Figure 1). The collection points were assigned to their respective regions - East, Central, and West. The state of Tennessee falls into three physiographic regions. The eastern third is part of the Appalachian Highlands. It is heavily forested and mountainous with marked parallel ridges and valleys. Agricultural land is limited to valley bottoms and roads tend to parallel the ridges. The central third of the state is part of the Interior plains and is much flatter. The density of agricultural land is higher although forest is still the dominant land cover in many of the counties. The road network is more uniform. The western third of the state is part of the Gulf - Atlantic Plain. Streams flow towards the Mississippi. The agricultural density of the land is higher in this region than the others. As in the central portion of the state, the road network is well developed but the Mississippi River poses a transportation barrier.

BRAVO Model Runs

Previously-developed county-level farmgate price and supply maps for the TVA region (Graham and Downing 1995, Downing and Graham 1995) were used as the biomass maps for the BRAVO runs. The potential wood chip supplies were assumed to be located at the centroid of the county. Each county had 14 potential supplies of wood chips each corresponding to one of the fourteen types of land that could be used to grow SRWC plantations.(4) The amount of each potential supply was based on the assumption that all the land with that farmgate price was used to grow SRWC plantations (e.g., assumed 100% farmer participation in SRWC production if the farmgate price was met). BRAVO was run at each collection point to generate a supply curve for that point. Only supplies from counties within 80 km (50 miles) of the collection point were eligible(5) The amount, farmgate price, transport cost and minimum delivered price (sum of the farmgate price and the transport cost) of each supply used in creation of the supply curve was extracted from the BRAVO run and entered into spreadsheets. Supply curves associated with 25% and 50% farmer participation rates were generated in the spreadsheet by reducing each individual supply amount (as extracted by BRAVO) by 75% or 50%.(6)

Cost and Demand Levels

The marginal costs of supplying the 10 quantities of wood chips listed in Table 1 for 100%, 50% and 25% farmer participation were extracted from the supply curves in the EXCEL spreadsheets. The mean transportation cost and farmgate price of the supplies needed to achieve a specific demand level were calculated for each location.

Once the marginal costs for a specific demand level were extracted from the curves for the 21 locations they were grouped by their region - east, central, west as were the mean transportation costs and farmgate prices. The regional means of those variables (6 to 8 locations per region) were calculated as were the standard error about those means. Plots showing the mean regional marginal costs, transportation costs and farmgate prices by demand were created. Maps were also made showing the marginal cost of chipped wood from SRWC plantations at each location for supplying a 100MW power plant and a 9,000 dry Mg/day (10,000 dry ton/day) ethanol plant.

Marginal Delivered Costs by Location and Demand

Figure 2 shows the mean marginal cost of supplying chipped wood in the three regions of Tennessee assuming 100% farmer participation - a best case. The central region is the least cost region unless very large amounts of woody chips are required (greater than 2.7 million Mg/yr). The eastern region is the highest cost region. Examination of the standard error reveals that the western region is the most homogeneous region; biomass supplies will cost about the same any place within this region regardless of demand. The eastern region shows considerable site to site variation in costs and this variation increases as demand increases. The central region appears intermediate in all regards. Figures 3 and 4 illustrate the point to point variability in feedstock costs at two demand levels.

Production versus Transportation Costs

Mean transportation and farmgate costs varied noticeably among the three regions (Figures 5 and 6).

The central region had the lowest transportation costs and the lowest farmgate costs until very high levels of wood chips were required (> 3 million dry Mg/yr). Farmgate costs were slightly higher in the western region than in the central region but transportation costs were markedly higher. The higher transportation costs in the west region can be attributed to the transportation barrier presented by the Mississippi River. The east region had the highest transportation costs averaging better than $15/dry delivered Mg ($13/dry ton) even at demands as low as 500,000 dry Mg per year and assuming 100% farmer participation. The high transportation costs can be attributed to both the mountainous terrain and the low density of agricultural land. Average farmgate prices were high in the east because the low density of crop and pasture land forced the use of land with high farmgate prices.

Impact of Farmer Participation Rates

Figures 7 and 8 illustrate the impact that farmer participation has on the marginal cost of supplying woody feedstock in the central and west regions.(8) The impact of farmer participation rate is much greater in the central region than in the west region. Assuming 100% farmer participation, the west region is the least cost region only for facilities demanding more than 2.7 million dry Mg/year (e.g. all but the very largest ethanol plant). Assuming 50% farmer participation, the west region is the least cost region for facilities demanding more than 1.3 million dry Mg per year (e.g. most ethanol plants) and some of the collection points in each region have an inadequate landbase (within the 80 km collection radius) for producing 6 million dry Mg (6.6 million dry tons) of wood a year. At a 25% participation rate, the western region is the least cost region for any plant requiring more than 750 million Mg per year (e.g. large power plants and all ethanol plants). Furthermore none of the collection sites in either region could support a 18,000 dry Mg per day ethanol plant without taking wood from counties farther than 80 km (50 miles) away and some sites in the central region could not support a 6,800 dry Mg day (7,500 dry ton day) ethanol plant. The within-region variability in delivered costs at a given demand level increases in the central region but not the western region as the farmer participation rate decreases. Thus from the perspective of risk reduction, all but the smallest-demand facilities would be better off sited in the western region than in the central region unless they had some mechanism to ensure high farmer participation rates.

Biomass facilities will always be faced with dilemma that their feedstock cost will be highly location specific. Using supply curves that are based on aggregate farmgate price information and uniform transportation costs can be misleading. Such curves may obscure opportune locations where biomass supplies can be had inexpensively while at the same time overestimating the total number of facilities that can be supported by feedstock under a specific price. Capturing the geographic complexity of potential biomass supplies is a necessity and one for which GIS is well-suited.

This study demonstrates one approach for quantifying the geographic complexity of biomass supplies and illustrates the need to consider the adoption rate of energy crops by farmers in projecting the likely cost of biomass feedstock.

Downing, M. and R.L. Graham. 1993. Evaluating a biomass resource: the TVA region-wide biomass resource assessment model. Vol I. pp 54-67. In Proceedings of the First Biomass Conference of the Americas. National Renewable Energy Laboratory, Golden CO.

Graham, R.L. and Downing, M. 1993. Renewable biomass energy: understanding regional scale environmental impacts. Vol III. pp 1566-1581. In Proceedings of the First Biomass Conference of the Americas. National Renewable Energy Laboratory, Golden CO.

Noon, C.E. 1993. TVA GIS-based biomass resource assessment. Vol. I. pp 74-78. In Proceedings of the First Biomass Conference of the Americas. National Renewable Energy Laboratory, Golden CO.

Downing, M. and R.L. Graham. 1995. The potential supply and cost of biomass from energy crops in the Tennessee Valley Authority Region. Biomass and Bioenergy. Accepted.

Graham, R.L. and M. Downing. 1995. Potential supply and cost of biomass from energy crops in the TVA region. Report ORNL-6858. Oak Ridge National Laboratory, Oak Ridge, TN.

1. Research sponsored by the Biofuels Systems Division, U.S. Department of Energy, under contract DE-AC05-840R21400 with Lockheed Martin Energy Systems, Inc.

2. The term "farmgate price" refers to the wood price needed to induce a farmer to grow SRWC plantations ($/Mg). In this study the farmgate price is assumed equivalent to the breakeven price - the wood price needed to assure the farmer a profit equal to his current profit. The supply amount associated with a specific farmgate price is the wood that could be grown if all the land of that farmgate price were put into SRWC production. Return to text

3. Profitability of crop land ranged from $0 to $450 per hectare ($0 to $182 per acre) after variable costs (e.g., seeds, fertilizer, farmer labor, custom harvesting), and fixed costs other than land were accounted for. The rental rate of pasture defined the profitability of pasture and ranged from $36.30 to $66.44 per hectare per year ($14.69 to $26.89 per acre per year). SRWC plantations were assumed to yield 5.4 to 9.6 dry Mg/ha/yr (2.4 to 4.3 dry tons/acre/yr) after harvest losses depending on the land type.Return to text

4. The crop and pasture land in the counties was divided into 14 classes which captured the range of soil suitability for growing conventional crops and/or SRWC. Each class had a unique SRWC yield and farmgate price.Return to text

5. Counties outside of Tennessee but within 80 km (50 miles) were included in the runs. To be included only some portion of the county needed to be within 80 km.Return to text

6. Halving participation or doubling demand will have the same effect on marginal cost of the supply. But as participation and demand are different concepts, it is useful to separate them.Return to text

7. The 18,000 Mg/day (20,000 tons/day) plant is larger than cellulosic ethanol plants are likely to be built. This value was included to show the shape of the curve. Cellulosic ethanol plants may be built that are smaller than 2,300 Mg/day (2,500 tons/day) but they are unlikely to be smaller than 450 Mg/day or 300,000 Mg/year (500 tons/day or 330,000 tons/yr) because of capital costs. Return to text

8. The east region shows the same trends as the central region but more dramatically. Most of the east sites cannot supply more than 2 million dry Mg regardless of delivered price because the total agricultural landbase is inadequate.Return to text