The authors are solely responsible for the content of this technical prentation. The technical prentation does not necessarily reflect the official position of the American Society of Agricultural and Biological Engineers (ASABE), and its printing and distribution does not constitute an endorment of views which may be expresd. Technical prentations are not subject to the formal peer review process by ASABE editorial committees; therefore, they are not to be prented as refereed publications. Citation of this work should state that it is from an ASABE meeting paper. EXAMPLE: Author's Last Name, Initials. 2010. Title of Prentation. ASABE Paper No. 10----. St. Joph, Mich.: ASABE. For information about curing permission to reprint or reproduce a technical prentation, plea contact ASABE at rutter@asabe or 269-429-0300 (2950 Niles Road, St. Joph, MI 49085-9659 USA).
An ASABE Meeting Prentation
Paper Number: 1009413
anticipateAn Optimal Storage and Transportation System for a
Cellulosic Ethanol Bio-energy Plant
Jason Judd, National Needs Fellow, Graduate Student
Industrial Systems Engineering Dept., Virginia Tech, Blacksburg, VA 24061
Subhash C. Sarin, Professor
Industrial Systems Engineering Dept., Virginia Tech, Blacksburg, VA 24061
John S. Cundiff, Professor
Biological Systems Engineering Dept., Virginia Tech, Blacksburg, VA 24061
Robert D. Grisso, Professor
Biological Systems Engineering Dept., Virginia Tech, Blacksburg, VA 24061
Written for prentation at the
2010 ASABE Annual International Meeting
Sponsored by ASABE
David L. Lawrence Convention Center
Pittsburgh, Pennsylvania
June 20 – June 23, 2010
Abstract . After locating a bio-energy plant for the conversion of biomass to biofuels, the logistics component associated with transporting the biomass is a nontrivial system. In this paper, we
propo local storage locations throughout an area for the temporary storage and loading of round bales. Prior results of uniformly distributing storage locations over an area are compared with tho obtained by using a mathematical programming-bad approach that eks optimal locations for the objective of minimizing the transportation and storage location costs. By having less storage
locations and longer haul distances, this approach achieves savings of 79%. The optimal solution us 39 storage locations compared with 167 locations propod in the previous study. With the transportation costs amounting to 65% of the total cost of biofuels, the savings are significant and show that, due to the complexity of the problem, human intuition and uniform storage placement
strategy may lead to results that are far from optimum. Instead, placement of more storage locations in areas with higher field densities (Mg/ha) provides a better solution. Moreover, such solutions are best obtained through the u of optimization-bad methodologies.
Keywords. biomass, switchgrass, biomass logistics, in-field hauling, hauling costs, satellite storage
locations, location allocation, integer programming, modeling
Background/Motivation
Over the cour of the last twenty years, there has been a new realm of thinking regarding the conversion of biomass to liquid fuel. Combustion engines are the driving force for the transportation of people or industrial goods in America. Cellulosic biofuels are fast becoming the most environmentally attractive and technologically feasible alternative to oil. With this new focus on biofuels, it is important to determine optimal solutions for the problems encountered in tting up a bi
o-energy plant and the accompanying logistics system.
Bio-energy has been sought-after as a means to reduce the greenhou effects on the environment caud by the u of oil as fuel. Many studies (including the well-known Billion Ton Study (Perlack et al., 2005)) have demonstrated the availability of potential biomass for conversion to energy (Downing and Graham, 1996). However, to make this alternative source of energy economically viable, it is esntial to minimize the total cost incurred from the farm gate to the gas pump. As much as 35%-60% of the total cost of a biofuel (ethanol) at the gas pump has been associated with transportation costs (Fales et al., 2008). Therefore, transportation costs constitute a significant portion of the total cost for this alternative source of energy. As rearch for better process to convert biomass to biofuel continues, larger plants with greater efficiencies will be built, which would also demand design of an effective logistical system. It is, therefore, necessary to ensure that an optimal system is designed for the transportation of biomass to the bio-energy plant. For many transportation models, due to the nature of established road networks, hauling by tractor-trailers is expected to be the most feasible and pre-established transportation method.
Regardless of the conversion technique ud at the bio-energy plant, the transportation problems associated with the generation of bio-energy are similar in nature. Much of this results from the fact t
hat raw biomass has a relatively low haul-density (Mg/m3). This results in multiple hauls across large coverage areas to satisfy the needs of the industrial-sized bio-energy plants, where haul volume is the limiting factor instead of weight capacities. Much work has been, and is being, done to densify the material as early on in the process as possible in order to increa the energy per load density. However, even in the prence of densification process, logistical issues will continue to persist that will demand achievement of lowest possible transportation costs. In their study, Yoshioka et al (2006) report that the tradeoffs between the cost to densify and, in turn, reduce transportation cost are significant enough to make densification a viable option. In Finland, a study was conducted concerning the densification of logging residues for u in a bio-energy system. Again, it was found that the savings resulting from densification outweighed the costs associated with the densification process (Ranta, 2005, Ranta and Rinne, 2006). Ravula et al. (2007, 2008) propo a round bale densification system with Satellite Storage Locations (SSLs). Moray et al. (2010) extend this work for a further densification process at the SSL. Also, work in this area has been reported by Cundiff (2007) and Ayoub et al. (2007).
Generic elements of the biomass logistics system
颐和园的英语There are certain elements that are common across all biomass logistics systems. The crop needs t
o be harvested and gathered within a given harvest window. This is followed by a densification process. Forage crops are often baled for densification propos and to make the material more manageable; trees are parated into trunks and the excess is baled; sugar cane and sorghum are chopped into a flowable material, and switchgrass is baled into round bales in the southeastern United States. This densified material is then stored on the production field or transported to a storage facility until it is needed for production. In some cas, this storage takes place at multiple non-central facilities (corn), while in other cas, storage is incurred on
each production field itlf (bales ud for cattle feed). After an unspecified amount of storage time, the material is converted into a new product and is transported as a usable source of bio-energy. This material, at the non-central storage facility, be it a production field or a condary storing point, is transported to a bio-energy plant. After the biomass has been converted to biofuel, it is stored until needed for demand via customer rvice centers or gas stations. Since the bio-energy plant has a continuous flow of product, both in and out of the bio-energy plant, this point can be ud to uncouple the transportation problem. In this paper, we focus only on the transportation of biomass from the production field to storage facilities.
This process has applications in almost all types of feedstock ud in the bio-energy industry. It is n
ecessary to design an optimal system such that the total cost is minimized. The items to be managed are the locations of storage facilities, the location of the bio-energy plant, and inventory levels at both the storage facilities and the bio-energy plant. The problem that we address in this paper is specific to Virginia’s switchgrass production for a bio-energy plant. In particular, we focus on determining optimal locations of the storage facilities.
Introduction and Problem Description研究生考试网
The implementation of a bio-energy plant to generate biofuel and/or electric power is a complex process and requires minimization of the cost incurred in order to reduce waste and maximize the profits for the company. The implementation of bio-energy plants will require shipment of large volumes of biomass from an area that is large enough to supply the bio-energy plant. A reduction in transportation costs will make a significant impact on the total cost of the system. In order to achieve this goal, we propo the u of mathematical models and methods to obtain optimal solutions of the logistic problems that ari in connection with the transportation of biomass from a local production field to an industrial size bio-energy plant.
Associated with this logistics system is an interconnected system of harvesting, storing, loading and
transporting the biomass. One of many critical issues in this regard is the high cost of transporting bales from a field by a dedicated and expensive piece of equipment. This equipment can be operated both in field conditions and on a paved road surface, the latter involving a constant interaction with traffic. Such equipment, typically a bale wagon, does not have very high transport speeds, thereby making its efficiency less than that of a tractor-trailer truck.
To help alleviate this problem, it is propod for each production field to store its biomass at a satellite storage location (SSL). The SSLs would be distributed throughout the area. Each SSL will cover either a single large production field or multiple smaller fields as propod by Cundiff et al. (2004). The purpo of a SSL is to provide a clo central location with a larger accumulation of biomass to enable the u of specialized loading equipment. The SSLs are only built in areas that have a high enough feedstock production density to ensure it is worthwhile to invest in the development of a SSL. Each SSL will be an uncovered gravel lot that is relatively clo to the main highways or paved (condary) roads. Table 1 provides 6 requirements for the SSLs. Requirements 2, 3 and 5 are absolute while 1, 4 and 6 may be violated (Resop et. al. 2010 (Hereafter, referred to as the “Resop study”)).
速写简单Each SSL will operate as a temporary storage location for biomass. Once enough biomass has bee
n accumulated, specialized loading equipment and trucks will be utilized to empty the SSLs. The biomass will be loaded onto trailers, transported to the bio-energy plant and either queued for immediate u or placed in at-plant storage for later u. This is an attractive system becau the SSL development and equipment will be relatively inexpensive while significantly reducing the cost of transportation through the u of mi trucks, which are a less expensive mode of hauling on a highway.
Table 1: A list of the six criteria ud for lecting and verifying sites for SSLs Criterion Description
1 At least 40 ha of potential production fields are available in a 3.
2 km radius
2 Must be on a state-maintained, condary road
3 Must be on non-forested land
4 Priority is given to scrubland or grassland
5 Site with an average slope in the range 0 to 10%
6 At least 0.4 ha of SSL area is available for each 40 ha of production fields
stored at the site
The cost to be minimized is that incurred for the transportation of the biomass from each production field to the nearest SSL. This will be achieved by minimizing the travel distance from each production field. An optimal system will provide for the loading and emptying of an individual SSL more than once during the year. Storage cost ($/Mg) decreas with each additional Mg stored per unit of capital cost incurred in building the SSL. Each SSL will be built on a suitable production field that meets the criteria of reducing its development cost.
In view of the above discussion, the problem that we address in this paper can be concily defined as follows: Given a t of production fields and their locations, determine the optimal number of SSLs, their locations (production fields on which to build the satellite facilities), and the allocation of production fields to SSLs so as to minimize the total cost of transporting biomass from production fields to the SSLs. We designate this problem as the biomass location allocation problem (BioLAP). In the next ction, we prent a mathematical formulation for this problem. This model is solved by a direct application of CPLEX, a well-known mathematical programming solver. This is followed by t
he description of a real-life scenario to which we apply our model. We also prent results of a comparative analysis of our results and tho obtained by using a heuristic procedure available for this instance in the literature.
Mathematical Formulation of the BioLAP
We u an integer programming (IP) formulation for the BioLAP that helps in conveniently capturing its various aspects. The model address the BioLAP at an aggregated (bi-annual) level. Thus, it is assumed that all the SSLs are full and need to be unloaded within a 6-month period. Also, the same yield rates are ud for each farm since the elements that effect yield rates (temperature, annual rainfall, growing ason, etc.) stay invariant throughout the area of study.
The propod model is a location-allocation model that choos the location of SSLs and assigns farms to each of the SSLs. Consider the following notation.
Parameters:
Cost associated with the u of a farm as a SSL
Cost of transporting biomass from farm to SSL
Cost of building SSL facilities
Variables:
1,if farm is lected as a SSL, 0,otherwi,∀ 1,…, .
1,
if farm us SSL as its SSL,0,otherwi,∀ , 1,…, .
We have the following formulation for the BioLAP.
Minimize: ,
Subject to:村居拼音版古诗
∀ , 1,…, .
1 ∀ 1,…,
0,1, 1,…, ; 0,1,
,
1,…, . The objective function consists of the cost of transporting biomass from production fields to the SSLs and the cost associated with the SSLs. The latter is divided into two components. The first of the, , pertains to the location where a facility is built, and the cond corresponds to the size of the facility. We assume that the cost per unit size of the facility is uniform across the area, and hence, given the size of the area, this cost can be determined a priori irrespective of where the SSLs are located. This is reprented by constraint t (1) captures the fact that a farm i will transport biomass to location j only if that location has been lected as a SSL.
Constraint t (2) ensures that a farm i can transport biomass to only one SSL. Constraint t
(3) constitutes logical requirements.
We solve this model to optimality by a direct application of CPLEX. It will be solved by first
taking linear relaxations of the problem, and then, by using heuristics and branch-and-bound/cut techniques to find an optimal solution. In this process, the linear problem (LP) is easy to solve, and it helps in curtailing unnecessary work for the branch-and-bound method. Additional constraints are added to the problem as need be until an optimal integer solution is found. Application to Real-life Scenario
In this ction, we provide a detailed description of a real-life scenario encountered in South Central Virginia to which the above model is applied. We u the databa developed in the Resop Study. This databa, referenced with permission, will hereafter be referred to as the “Resop Databa”. In that study, 167 SSLs were located within a 48 km radius of Keysville, Virginia. Our objective is to determine optimal locations of the SSLs, the assignments of
production fields to SSLs, and compare our results with tho prented in the Resop study in which they provided a somewhat uniform distribution of SSLs throughout the study area. In the Resop study, the SSLs were created one-at-a-time by creating 3.2 km buffer rings around
potential SSLs. This process was not the focus of Resop’s paper and we do not intend to infer problems with the analysis done in that study. It was known at the time that the locations were not optimal, but assumed that they were good enough for the study that was conducted. In this paper, we provide a glimp of the level of improvements that can be achieved over heuristic solutions by using optimization-bad methods for such problems. (1) (3) (2)
The relevant overview of the databa is shown in Table 2. As may be en in row 1 of Table 2, the data initially contained 24,000 farms. Through data reduction techniques, described in the next cti
on, the were merged into 2,348 farms. The total farm area is large enough that even with lower than expected production yields of switchgrass, a bio-energy plant could effectively operate within 3-5% of an optimal plant size (Searcy et al., 2007).
表格插入图片Table 2: Databa Information Original number of farms
医院中午休息吗24,000 Reduced number of farms
2,300 Number of potential SSLs有关理想的名言警句
594 Area (ha)
Total 68,529 Maximum 528.2 Average 30.3 Minimum 6.07
Databa Management
In the Resop databa, data was collected to find the land that could potentially be attracted into the production of switchgrass. Farms were lected to be SSLs by lecting a starting SSL with a 3.2 km radius in ArcGIS, and then, physically lecting the SSL locations to cover most of the study area in a somewhat uniform fashion using 3.2 km radius rings. Farms that were more than 10 km from an
y potential SSL are not expected to be allocated to a potential SSL, and therefore, were removed from the data. We estimate that the number of production fields
attracted into producing switchgrass will be 50% of the initial farms. Although this assumption is very conrvative, it provides the initial structure for the transportation network and the
establishment of SSLs. If additional farms are attracted into production, they may be added directly into the current solution and considered for transporting their switchgrass to the nearest pre-established SSL.
Figure 1: Display of map layers
