UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Simulation modeling of forest biomass operations and harvest residue moisture content Pledger, Sean 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_may_pledger_sean.pdf [ 2.06MB ]
Metadata
JSON: 24-1.0300235.json
JSON-LD: 24-1.0300235-ld.json
RDF/XML (Pretty): 24-1.0300235-rdf.xml
RDF/JSON: 24-1.0300235-rdf.json
Turtle: 24-1.0300235-turtle.txt
N-Triples: 24-1.0300235-rdf-ntriples.txt
Original Record: 24-1.0300235-source.json
Full Text
24-1.0300235-fulltext.txt
Citation
24-1.0300235.ris

Full Text

SIMULATION MODELING OF FOREST BIOMASS OPERATIONS AND HARVEST RESIDUE MOISTURE CONTENT By  Sean Pledger  B.S.F, The University of British Columbia, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   April 2016   © Sean Pledger, 2016  ii  Abstract In order to limit the effects of anthropogenic climate change the world is moving away from the use of fossil fuels as a primary energy source. Bioenergy is expected to form a substantial contribution to this transitional strategy. In order to increase bioenergy production, underutilized forest harvest residues are being targeted as a fuel source. Even with favorable policies in place to encourage their use, the processing and collection of these previously disregarded resources is often prohibitively expensive. Quality factors such as material moisture content also impact the viability of harvest residues for fuel purposes.  As a result, careful operational planning is of great importance to sourcing high quality, economically feasible biomass.  To gain a better understanding of the forest biomass supply chain, a simulation model was developed for a case study located in coastal British Columbia, Canada. A seasonal moisture content trend was identified and incorporated to help develop a strategy for sourcing high quality materials. It was found for BC’s coastal temperate rainforest environment that by delaying biomass collection until the second summer after timber harvest an average delivered moisture content of 28% can be achieved rather than 38% is operations proceed in the first summer. This reduction in delivered moisture content also led to a decrease in delivered cost from $72.08 to $67.95 per oven dried tonne.  Trucking and equipment configurations were also examined to identify least cost approaches to biomass collection under varying conditions. Comparing high productivity and low productivity equipment configurations showed a $26.08/ODT cost increase when switching to less productive equipment. By employing an electric centralized grinder transporting unprocessed harvest residues, costs were shown to decrease for all cutblock groups with a cycle time of less than four and a half hours. Least cost fleet size was found to be largely dependent on the average cycle iii  time to the biomass source. And the volume of available biomass at a given cutblock was found to have an impact on delivered costs with a 20% increase in biomass volume resulting in a cost decrease of greater than $2/ODT.    iv  Preface Dr. Gary Bull initiated the study and ensured its support. Dr. Saeed Ghafghazi developed the AROMA simulation modeling platform and contributed significantly to the case study model development. Dr. Ghafghazi also provided invaluable advice in structuring the thesis. Data inputs and an in-depth understanding of the case study system were shared by Ian McIver. I developed the visual basic analysis tool, helped develop then calibrated the case study model, analyzed the model results, and drafted the thesis.    v  Table of Contents  Abstract .............................................................................................................................. ii Preface ............................................................................................................................... iv Table of Contents ...............................................................................................................v List of Tables .................................................................................................................. viii List of Figures ................................................................................................................... ix List of Equations ................................................................................................................x List of Abbreviations ....................................................................................................... xi Acknowledgements ......................................................................................................... xii Dedication ....................................................................................................................... xiii 1 Introduction ................................................................................................................1 1.1 Research Objectives ............................................................................................ 5 1.2 Modeling: Simulation and Optimization ............................................................ 5 1.3 Supply Chain Modeling ...................................................................................... 7 1.4 Logistics Modeling ........................................................................................... 10 1.5 Forestry & Biomass Modeling .......................................................................... 11 1.6 Moisture Content .............................................................................................. 19 2 Methods .....................................................................................................................24 2.1 AROMA Simulation Model .............................................................................. 24 2.2 Simulation Output Processing and Analysis Tool ............................................ 25 3 Study Area Description ...........................................................................................26 3.1 Climate and Forest Type ................................................................................... 26 vi  3.2 Infrastructure and Forest Operations ................................................................ 27 4 Input Parameters .....................................................................................................30 4.1 Geospatial network ........................................................................................... 30 4.1.1 Road Accessibility ........................................................................................ 31 4.1.2 Supply Attributes .......................................................................................... 34 4.2 Equipment & Costs ........................................................................................... 37 4.3 System Behavior: Schedules and Delays .......................................................... 40 4.4 Moisture Content .............................................................................................. 44 5 Results .......................................................................................................................46 5.1 Fixed Fleet Planning ......................................................................................... 46 5.2 Alternate Equipment ......................................................................................... 48 5.3 Variable Fleet Cutblock Grouping .................................................................... 51 5.4 Centralized Electric Grinding ........................................................................... 56 5.5 Sensitivity Analysis .......................................................................................... 60 5.5.1 Biomass Volume ........................................................................................... 60 5.5.2 Delivered Moisture Content .......................................................................... 62 5.6 Comparison of Methods .................................................................................... 65 6 Discussion..................................................................................................................67 6.1 Sensitivity in Aggregate .................................................................................... 67 6.2 Varying Complexity of Methods ...................................................................... 73 7 Conclusion ................................................................................................................75 7.1 Reducing Costs ................................................................................................. 75 7.2 Moisture Content .............................................................................................. 76 vii  7.3 Benefits of Simulation ...................................................................................... 77 7.4 Future Work ...................................................................................................... 78 Bibliography .....................................................................................................................79  viii  List of Tables Table 1 Biogeoclimatic Zones of Study Area ............................................................................... 26 Table 2 Simulation Model Network and Object Attributes .......................................................... 33 Table 3 Road Network Speed Classes .......................................................................................... 33 Table 4 Biomass Ratios (Macdonald, 2012) ................................................................................. 35 Table 5 Average Oven Dried Weight of Local Wood Species ..................................................... 35 Table 6 Equipment Hourly Costs and Capacities With and Without Fuel Consumption ............. 38 Table 7 Harvest Residue Moisture Content Sampling Results on Vancouver Island (Dyson, 2013) ............................................................................................................................................. 44 Table 8 Least Cost Fleet Size and Costs for Fixed Fleet Configurations ..................................... 50 Table 9 Delivered Cost of Biomass for Different Equipment Types by Block Group ................. 54 Table 10 Delivered Cost Breakdown of Operations Under Various Equipment Configurations . 58 Table 11 Delivered Cost by Block Group Under Various Equipment Configurations ................ 59 Table 12 Delivered Cost Sensitivity to Biomass Volume ............................................................ 61 Table 13 Delivered Cost and Operating Days In Response to Various Moisture Content Schedules ...................................................................................................................................... 64 Table 14 Cost Breakdown of Operations by Simulation and Input/Output Model ...................... 66  ix  List of Figures Figure 1 Map of Biogeoclimatic Zones in Study Area ................................................................. 27 Figure 2 Map of Study Area's Road Infrastructure ....................................................................... 29 Figure 3 Map of Road Network .................................................................................................... 36 Figure 4 Anticipated Seasonal Moisture Content Variation of Harvest Residues on BC's Coast (Acuna et al., 2012; Dyson, 2013) ................................................................................................ 45 Figure 5 Delivered Cost of Biomass with High Productivity Diesel Equipment ......................... 47 Figure 6 Cost Breakdown of Biomass Delivery for HP Equipment and 1 Operation .................. 48 Figure 7 Delivered Cost of Biomass with Low Productivity and High Productivity Diesel Equipment ..................................................................................................................................... 50 Figure 8 Map of Cutblock Groups ................................................................................................ 52 Figure 9 Least Cost Fleet Size Charts by Block Group ................................................................ 53 Figure 10 Impact of Block Group Cycle Time on Delivered Cost ............................................... 55 Figure 11 Impact of Block Group Cycle Time on Least Cost Fleet Size for HP Equipment ....... 55 Figure 12 Cost Impact of Equipment Productivity and Cycle Time by Block Group .................. 56 Figure 13 Delivered Cost of Biomass with Centralized Grinding ................................................ 57 Figure 14 Delivered Cost of Biomass Comparison between Electric Centralized Grinding and High Productivity Diesel............................................................................................................... 58 Figure 15 Cost Difference Between HP Diesel and CE Configurations by Block Group ............ 60 Figure 16 Cost Sensitivity to Biomass Volume Relative to Fleet Size ......................................... 62 Figure 17 Moisture Content Input Schedules ............................................................................... 63 Figure 18 Cost of Delivered Biomass at Various Average Moisture Contents ............................ 64 Figure 19 Operating Days Required to Deliver Biomass at Various Average Moisture Contents65 x  List of Equations Equation 1 Moisture Content of Wood ......................................................................................... 20    xi  List of Abbreviations  BC  British Columbia BCTS  British Columbia Timber Sales CE  Centralized Electric Grinding Equipment Scenario GIS  Geographic Information System HBS  Harvest Billing System HP  High Productivity Equipment Scenario HSPP  Howe Sound Pulp and Paper I/O  Input / Output  IBSAL  Integrated Biomass Supply Analysis and Logistics Model LCA  Life Cycle Analysis LP  Low Productivity Equipment Scenario MC  Moisture Content MKRF  Malcolm Knapp Research Forest NPV   Net Present Value ODT  Oven Dried Tonne VU  Volumetric Unit xii  Acknowledgements I would like to thank my supervisors Dr. Gary Bull and Dr. Saeed Ghafghazi for their continuous support and patience in completing this analysis and thesis. Gary, for the endless opportunities he provided in developing industry contacts and employment opportunities, as well as for coaching me in methods of economic analysis and management. Saeed’s training in programing and advanced modeling was pivotal in the completion of this work. I offer my gratitude to Dr. John Nelson and Dr. Taraneh Sowlati for providing guidance at the outset of this research, as well as to PhD Candidate Kyle Lochhead for always being available to clarify principals of scientific research and modeling. I am grateful for the financial support from Mitacs, the VCO network, Genome Canada and Ledcor Resources and Transport, without whom this research would never have been completed. Finally, I would like to thank my parents and my partner Jaclyn for their extreme patience and unwavering support over the years that it has taken to complete this chapter of my life.  xiii  Dedication To my parents, Kathy Nomme and Wayne Pledger.  And to my partner in all of life’s adventures, Jaclyn Lord-Purcell.     1  1 Introduction The world is moving away from reliance on fossil fuels as an energy source (United Nations, 2015). This involves developing new renewable energy system and transitioning off greenhouse gas intensive fuels. Bioenergy forms a substantial contribution to this strategy (Eisentraut & Brown, 2012). For Canada’s forest industry this global issue represents both a threat from changing ecosystems and lengthening wild fire seasons, as well as an opportunity to utilize undervalued resources. In 2008 British Columbia implemented the BC Bioenergy strategy. The overall aims of this program were to ―create new opportunities for rural communities, spur investment and innovation, and help BC reach the goal of energy independence by 2016‖ (BC Office of the Premier, 2008). In the years that have followed, research and development of BC’s bioenergy industry has grown slowly, with significant focus put on utilizing unmerchantable mountain pine beetle trees. This focus resulted in limited investigations of BC’s coastal biomass industry, where both climate and logistics systems vary greatly from the provinces interior. Internationally, using agriculture and forest residues to generate energy has been a growing field of study for decades. A range of plant species have been identified whose bone dry energy densities range from 17 to 21 MJ/kg which makes them suitable as a bioenergy feedstock (McKendry, 2002). Research has been conducted on most aspects of forest harvest residues for bioenergy, from operations and logistics through to the plant properties and the ecological impacts of their removal (Hakkila, 1989). There are a number of technologies used to convert biomass into energy. The combustion of biomass to generate steam and spin a turbine to create electricity is referred to as a biomass 2  boiler system. Whereas the anaerobic heating of biomass to produce a mixture of volatile organic compounds, known as syngas, to power an engine is referred to as a biomass gasification system (USEPA, 2007). These technologies have been extensively utilized in recent history by a range of industries including pulp and paper facilities. As demand for bioenergy increases, so does the research and development in generation technologies. A common problem identified in commercializing these systems for utility scale energy is the high cost of biomass collection and transportation (Son, Yoon, Kim, & Lee, 2011). In addition to generating heat and electricity from biomass, research and development is ongoing in the field of biofuels which propose to replace or partially substitute petroleum based auto fuels and chemicals (Naik, Goud, Rout, & Dalai, 2010). With favorable policies and technological advancements the use of wood pellets has grown globally to generate both heat and electricity at the residential and utility scales. It has been found that forest harvest residues may be a valuable feedstock source for producing these pellets (Lehtikangas, 2001). Potential issues have been raised over using forest harvest residues for bioenergy purposes. These include concerns over the loss of soil nutrients through biomass removal, and speculation regarding the climate benefit associated with bioenergy production. In BC, harvest residues amount to an estimated 15.5 megatonnes per year (oven dried basis) and have historically been piled and burnt on site (Paré, Bernier, Thiffault, & Titus, 2011).  When comparing bioenergy production from harvest residues to the status quo of open air burning, the controlled combustion of material at a facility has been found to be more efficient. This can result in a 39% reduction in CO2 emissions and an 89% reduction in particulates (Jones, Loeffler, Butler, & Hummel, 2009). When considering emissions from processing and transportation, as 3  well as the offsetting of fossil fuel energy, Jones et al. identified a 43% reduction in emissions when compared to pile burning. Numerous trials have been conducted to investigate the impacts of biomass removal on soil nutrients and forest productivity. Fleming et al. (2006) found that whole tree harvesting had no significant impact on stand volume index after 5 years. Different forest types will be impacted in different ways but in general a majority of the ―nutrient capital‖ of forest ecosystems is held in below ground pools and the removal of biomass residues ―should result in only a small percentage of nutrient loss‖ (Farve & Napper, 2009). It was identified that an atmospheric input of Nitrogen between successive harvests was sufficient replenishment. It has however been noted that Calcium depletion could be of concern in some hardwood forests. In general ―responsible biomass removal can be done without noticeable soil or site quality impacts‖ (Evans, 2008). Although the collection and transportation of forest harvest residues has been ongoing for years, the economics of these operations continue to pose a significant hurdle. In examining over 30,000 ha of forest land for biomass removal, Jones et al. (2009) found that only half of the sites could be processed economically. This is the result of the highly variable costs and revenues associated with these complex supply chains. Where the available volume of feedstock is of great importance in managing operations, estimating it also tends to be highly variable given available information. Esteban and Carrasco (2011) for example, estimated the available biomass across Europe based on information about tree diameters, heights and densities.  There are numerous options for transporting forest harvest residues, including chipped hog fuel, loose residues, and residue bundles. Residue bundles, also known as compact residue logs, have been identified as an efficient form of outdoor winter storage in mountainous climates such as the Italian Alps (Spinelli & Magagnotti, 2009).  However they increase costs in the forest 4  harvest residue supply chain through additional handling. A further benefit of bundled residues is the ability to transport them using the same equipment as conventional roundwood harvesting in Europe. While their use in North America’s biomass industry is relatively absent, operations in Europe have employed this method for years. Johnansson et al. (2006) investigated the pros and cons of transporting bundles versus chips and found costs to be comparable for short to medium transport distances. Whereas other studies have found slash bundling to be the most economical option for travel distances beyond 60km (Kärhä & Vartiamäki, 2006).  This goes to show that the myriad of operational decisions required in running a forest biomass supply chain are complex and can greatly impact the profitability of the business. With an extensive body of research on forest biomass supply chains there have been numerous learning’s. Decades of experience in Europe have resulted in improvements in equipment, changes in organizational structures, and other operational efficiencies which have led to cost reductions (Junginger, Faaij, Björheden, & Turkenburg, 2005). Meanwhile the perception of industry managers going forward is an expectation that the practice of roadside chipping will decrease and processing at terminals or bioenergy plants will increase (Kärhä, 2011).  In North America delivered forest residue costs have been estimated to range from around $50 to over $150 per oven dried tonne (Stennes, McBeath, & Centre, 2006). Equipment designed for high productivity tends to be very fuel intensive and costly, with grinding and loading operations in some cases consuming 140 liters of diesel an hour (Smith, 2010). A predominant issue with the forest harvest residue supply chain is that no two operations are the same. Variations in geography, policy, and equipment availability limit the options available to managers. In order to conduct efficient and profitable operations managers need information and tools to assist their decision making.  5  1.1 Research Objectives In order to increase the supply of biomass energy and offset the use of fossil fuels, the energy and forest industries of British Columbia are faced with the challenge of sourcing previously discarded forest biomass from cutblocks. Biomass has been sourced more extensively from mountain pine beetle killed stands in the provinces interior region, while the coast has been largely avoided due to its challenging terrain and significantly higher annual precipitation. The collection and transportation of logging residue can be a prohibitively expensive process, even with subsidized biomass power generation. Consequently it is important to understand the operational costs of delivering this material. Residues sourced from the interior of the province are typically also lower in moisture content than those on the coast, making them more desirable as a fuel source. The first objective of this research is to analyze the costs of delivering forest biomass to a central location and determine how they will be impacted by adjusting operational configurations. The second objective of this research is to identify any seasonal patterns in harvest residue moisture content that may exist on the BC coast; then devise a management strategy to improve the delivered quality of biomass.  1.2 Modeling: Simulation and Optimization The study of industrial systems and the processes within them is nothing new. With its birth in military strategy, the science of industrial dynamics was formalized with the advent of digital computers following the Second World War. This applied the simulation and optimization techniques used in code breaking and troop deployment to the supply chains of commercial industry for the purpose of improving efficiency. Model experimentation is used to not just study 6  parts of a system but the interactions between these parts in order to identify unexpected and sometimes troublesome phenomenon (Forrester, 1961). Simulation modeling is the process of designing and experimenting on a model of a real system for the purpose of understanding the system’s behavior and evaluating strategies for operating the system (Shannon, 1998). As a mathematical process for describing systems, simulation modeling is characterized by a series of steps which vary based on the type of analysis being conducted. Law (2009) outlines the  general steps of developing a simulation model beginning with problem formulation, then data collection, generating and testing assumptions, programming and checking validity, and finally experimentation. In contrast to this guide for simulation modeling, Kleijnen (1997) discusses an approach to evaluating simulation models. This begins with analyzing the validity of the model by comparing it to the real world system, and then follows with screening of inputs, sensitivity analysis, uncertainty or risk analysis, and its use of optimization. Simulation and its applications are as varied as the systems of the physical world. At the tangible end of the spectrum simulation can be used to analyze systems with physical actors such as fleets of fishing vessels to understand policy impacts (Soulié & Thébaud, 2006). Similarly models can be created to study the interface between ecological and social systems to understand the trade-offs made by society in allowing varying levels of resource extraction (Bousquet, Barreteau, Le Page, Mullon, & Weber, 1999). While more abstract systems can also be studied through simulation, such as deregulated electricity markets for the purpose of predicting market evolution and the behavior of the participants (Praça, Ramos, Vale, & Cordeiro, 2003). With the range of applications for simulation there are also a range of platforms that have been developed 7  such as SeSAm (Klügl, Herrler, & Fehler, 2006) and JaamSim (King & Harrison, 2013a) to assist analysts and researchers with model development. Optimization models are, most commonly, methods of mathematical programming used to find an optimal outcome to a stated problem. These methods include linear programming, mixed integer programming, dynamic programing, as well as other variations and combinations of modeling principals (Joseph Buongiorno, 2003). Optimization models can be powerful tools to solve clearly defined questions. Their applications range from pure computing and mathematics, to problems in business management, and emergency response logistics and disaster planning (Caunhye, Nie, & Pokharel, 2012).  Simulation and optimization methods can be combined or used in isolation. Optimization is often used for resource allocation problems, but when these problems are dynamic with distributed information and decision making, simulation methods are found to be complementary (Davidsson, Johansson, Persson, & Wernstedt, 2003). There are a variety of methods for combining simulation and optimization techniques depending on the problem at hand. To optimize stochastic simulations for example, a Simulation Optimization approach has been proposed where input parameters are tested until the target objective is minimized (Amaran, Sahinidis, Sharda, & Bury, 2014). Similar approaches have been identified for use in supply chain management where iterative simulations are run to identify causal relationships between key decision variables and supply chain performance (Wan, Pekny, & Reklaitis, 2005). 1.3 Supply Chain Modeling Manufacturing business processes can generally be grouped into 5 stages: Supplier, Manufacturer, Distributor, Retailer, and Customer. This process flow represents the basic supply chain of any product (Chopra & Meindl, 2010). With products of different complexity, these 8  stages and the resulting supply chain can look quite different.  For a computer manufacturer a supplier of microchips will have its own complex supply chain which will differ significantly from a log supplier to a pulp producer.   Supply chains can be modeled and analyzed using a variety of methods, with simulation and optimization approaches being popular among researchers and analysts.  While there are some commonalities between these processes, when mathematical programming is employed the analysis is often predominantly optimization. Alternatively, simulation modeling involves observing and analyzing a system over time. Some supply chain models use these approaches in isolation or in combination, while others resemble more simplistic input/output models. The use of computer models as tools to improve the efficiency of manufacturing and distribution systems has accelerated along with the computing power that has enabled it (Shen, 2006). As supply chains are often complex systems which can be analyzed at varying scales, different approaches are often used to study them at the strategic level as compared to the operational or tactical level (Tako & Robinson, 2012). Supply chains can be viewed as a flow of materials through the various stages, where at each stage there is an inventory of one or multiple materials or products. In a Push/Pull view of supply chains the inventory fluctuations within the supply chain are dived into two categories: Push if they are triggered by the supplier, or Pull if they are triggered by demand (Chopra & Meindl, 2010). The use of simulation in studying supply chains can be used beyond monitoring material flows. The potential for What-If analysis allows a much broader contemplation of a system. Where a single model can be used to identify phenomenon at an operations level such as ―inventory costs being lower at lower levels of added value‖, the same analysis can be used to recommend company acquisitions at a strategic level (Towill, Naim, & Wikner, 1993).  Various 9  metrics can also be used to analyze simulated systems, with cost or Net Present Value (NPV) of supply chain operations being common.  Amini et. al. (2012) used NPV to study not only supply chain operations, but also consumer behavior and the effectiveness of marketing activities. This analysis was used to determine profitable sales production policies. Simulation models have also been used to study the impacts climate change may have on supply chain transportation logistics by incorporating weather factors and delays (Maoh, Kanaroglou, & Woudsma, 2008). Modeling supply chains can become an incredibly complex task depending on the level of detail included, and the scope of the network and objects. With the large number of assumptions that go into these models it has been found that some mathematical and operation research approaches do not perform well in isolation, due to the stochastic nature and high uncertainty in supply chains (Stefanovic, Stefanovic, & Radenkovic, 2009). While optimization models due exist which are designed to handle uncertainty, using simulation to address ―What if‖ questions within existing or theoretical supply chains can also alleviate some of these concerns. Furthermore, mathematical optimization models can be incorporated into simulation models to identify optimal decisions in selecting suppliers or identifying target markets  (Akanle & Zhang, 2008). Object oriented programming is language that has used to develop supply chain simulation models, with a variety of frameworks being proposed (Rossetti & Chan, 2003). Along with the frameworks and methodologies, there are also a range of programming languages (Alfieri & Brandimarte, 1997) and software packages (Chatfield, Harrison, & Hayya, 2003) designed to help users develop their own models. Object oriented programming generic environment for developing simulation models, with fewer limitations than off-the-shelf 10  platforms. This can result in customizable decision support tools which can incorporating optimization techniques throughout the supply chain (Biswas & Narahari, 2003) Alternatively supply chain models can be predominantly optimization based. These models are commonly used in planning facility locations and broader network designs (Klose & Drexl, 2005). Optimization is also widely used in supplier selection within supply chains (Ghodsypour & O’Brien, 2001). Given a variety of tools to study and improve supply chains it is important to consider that efficiencies can be found with simple process changes such as increasing communication with the business (Carlsson & Rönnqvist, 2005), and that building these tools are just part of the analytical process. 1.4 Logistics Modeling Within the field of supply chain modeling there are a number of different focal points, such as transportation logistics. This field focuses on reducing costs associated with the physical transportation of materials. Vehicle routing problems concern the routing, scheduling and fleet planning of a transportation force, and are a common approach to generating efficiencies in forest products supply chains (Audy & Rönnqvist, 2012). The traditional approach to identifying logistics strategies is by way of linear programing to quickly generate and evaluate a large number of network designs. However, these models are not readily able to consider stochastic events. As a result simulation has become a popular tool in the analysis of transportation systems (Caputo, Pelagagge, & Scacchia, 2003). As with the broader field of supply chain analysis, numerous frameworks have been proposed to assist with the development of logistics simulation models (Manuj, Mentzer, & Bowers, 2009). These frameworks differ in a number of ways, but can generally be categorized as either discrete event or agent based simulation. While these methods are not necessarily 11  mutually exclusive, each has its advantages depending on the problem at hand (Becker et al., 2006). Simulation has been touted as a powerful tool in the analysis of complex logistics systems due to its ability to emulate systems in great detail (Lehtonen & Seppala, 1997). While simulation’s value as a decision support tool is well documented, there is also significant value gained during model development. Problem formulation and data collection, followed by calibration and analysis enable a modeler to gain an intuitive and detailed understanding of the logistics system they are attempting to recreate (Chen, Lee, & Selikson, 2002). As with the broader field of supply chain modeling numerous platforms have been developed to assist with building logistics models, such as the multiagent based simulation system PlaSMA (Gehrke & Ober-Blöbaum, 2007). Optimization approaches also provide valuable tools to logistics modeling. As with the birth of systems dynamics in military applications, linear programing methods are still employed by naval forces to optimize fleet deployment and resupply (Brown, 2006).  One of the most common uses of optimization in logistics planning is in identifying facility or depot locations to improve supply chain efficiency (Bosona & Gebresenbet, 2011). 1.5 Forestry & Biomass Modeling Supply chain modeling has been used to gain insight and identify efficiencies in commercial, industrial and military systems across the globe. Its use in the forest products industry stretches back decades. Now, with a global push to increase utilization of natural resources and the development and expansion of bioenergy industries, supply chain modeling is finding new applications.  12  Bioenergy and biofuel supply chains include a wide spectrum of raw material inputs. These range from animal manure and corn stocks as part of agricultural biomass industry, to urban wood waste and forest harvest residues for wood based biomass (An, Wilhelm, & Searcy, 2011). Each specific bioenergy generation technology has its own unique supply chain, and every geographic region has its own complex issues to overcome. The United States has been a major supporter of biomass energy, in part to reduce carbon emissions but also to limit the countries dependence on imported oil. Supply chain simulation modeling has been shown to be a valuable tool not only in improving operational efficiency and reducing costs but also in estimating the emissions associated with procuring biomass for energy purposes (Zhang, Johnson, & Johnson, 2012). A prominent simulation model used for analyzing agricultural biomass supply chains is the integrated biomass supply analysis and logistics model (IBSAL). Developed in 2006 by Sokhansanj, Kumar and Turholloh, IBSAL is a discrete event simulation model that uses working rates and queues to represent the material flow network of a supply chain. Of particular value is IBSAL’s ability to account for quality factors of biomass including moisture content (MC) and dry matter loss within the system (Shahab Sokhansanj, Kumar, & Turhollow, 2006). Sokhansanj et al. (2008) used IBSAL to compare harvesting and bailing systems of straw within the biomass energy supply chain to investigate the costs, energy input, and emissions associated with each treatment.  In 2011 Ebadian et al. expanded IBSAL to include multiple feedstock sources. It was identified that a proposed biomass ethanol plant had sufficient annual supply to operate but could only meet daily demand during part of the year. The model was used to determine storage needs and costs to enable the year round supply of biomass to the proposed plant (Ebadian, Sowlati, Sokhansanj, Stumborg, & Townley-Smith, 2011).  The model was further developed to include 13  an optimization function identifying which farms should deliver biomass to the ethanol plant, and allocating farms to specific storage facilities. Once the optimal network was determined through the mathematical programing technique, the simulation component of IBSAL was run to better evaluate the costs associated with equipment operations (Ebadian, Sowlati, Sokhansanj, Townley-Smith, & Stumborg, 2013). As with its application in IBSAL, optimization is largely used in network design of biomass supply chains. Dal-Mas et al used a similar approach to design an optimal supply network for an ethanol plant, but conducted their analysis from the perspective of minimizing financial risk (Dal-Mas, Giarola, Zamboni, & Bezzo, 2011).  Forestry and forest biomass supply chains differ from those of agricultural biomass in the sense that logs and harvest residues are not taken from the same place twice within a business cycle. A geographical region may be a continuous source of resources but every harvest will be in a slightly different location, with the potential need to develop new road infrastructure and face different operational challenges. As a result, tools have been specifically developed for the forest products manufacturing industry. Frayret et al. for example, developed a generic architecture for advanced planning and scheduling of forest products supply chains using agent based simulation (Frayret, D’Amours, Rousseau, Harvey, & Gaudreault, 2007). IBSAL has also been adapted for use in the forest biomass industry. Mobini et al. (2011) used this framework to evaluate the costs of delivering biomass to a proposed wood based bioenergy facility in Quesnel, BC. This analysis also included an assessment of the MC of delivered material based on environmental factors.  Simulation modeling of forest operations has been used for decades to identify efficiencies in supply chains. In coastal environments barging of logs or residues adds complexity to these operations. By developing a simulation model of logging operations on 14  islands in Finland, Asikainen (2001) identified that using multiple barges would increase the utilization of more costly equipment, thereby reducing overall delivered log costs. Similarly, it was found through simulation modeling that in the case of stump crushing and recovery for bioenergy, as the travel distance increased, additional trucks were required to keep costs down via increased utilization of crushing equipment (Asikainen, 2010). Further development of IBSAL to include British Columbia’s wood pellet supply chain was completed to analyze the cost and emissions associated with the industry (Mobini, Sowlati, & Sokhansanj, 2013). Additional analysis was then conducted on pellet distribution to European markets, where it was identified that the additional capital investment in torrefaction plants was warranted to increase the energy density of the fuel, thereby reducing the transportation costs per unit of energy potential (Mobini et al., 2014). Mobini’s work included sawlog harvest scheduling and additional supply analysis for biomass power facilities. This included a MC analysis of biomass. Within the model MC was calculated based on historic ambient weather conditions with a fixed value used in a simulation run. Sensitivity analysis was conducted around that value (Mobini, 2015). In studying New Zealand’s forest biomass industry through simulation it was identified that the simplest collection systems tended to be the cheapest, as increased handling lead to increase costs (Hall, Gigler, & Sims, 2001). The same study also found the model was particularly sensitive to MC, material density, and machine data inputs, and emphasized the importance of reliable data. Simulation has also been used to identify an optimal harvesting, processing and transportation strategy for delivering short rotation eucalyptus biomass to a power facility in NZ. Here the authors compared intensively harvesting and storing the entire annual supply in a short period with slower less intensive operations and minimal storage (Sims 15  & Venturi, 2004). It was found that the additional handling associated with longer term storage was unproductive. This analysis also considered the biomass MC.  It was assumed that green residues had a MC of 60% and after one month of field storage that would drop to 45%. To study Finland’s woody biomass supply chain Tahvanainen and Anttila (2011) developed a simulation model to examine several methods of forest harvest residue collection. These options included the use of trucks, barges and railways, as well as loose residue transport, on site grinding and residue bundling. The authors were able to identify that for short distances (less than 60 km) transporting unprocessed residues was the most economical. Beyond 60 km on site grinding was the most cost effective up to 135 km, beyond which point the use of railways could be justified. In their case study the use of residue bundles and waterways was not competitive at any distance.  The use of optimization in forestry supply chains is extensive, particularly with regards to log bucking, truck scheduling, and production planning (Rönnqvist, 2003). With regard to biomass supply chain planning, a common use of optimization is in facility and storage depot determination. In the US methodologies have been developed to take advantage of  the countries high resolution forest inventory and price data to locate bioenergy and biofuel plants (Stasko et al., 2011). In Ontario, Canada optimization methods were used in allocating forest harvest residues between 4 power plants (M. Alam, Pulkki, & Shahi, 2012).  Cambero et al. (2015) studied the forest biomass supply chain in BC and used optimization to identify a regional biomass facility type and location using NPV as a metric. Gan (2007) used optimization to determine the capacities of biomass power plants given the available supply of logging residues in the U.S.. Palander and Voultilainen (2013) studied the impacts of roadside terminals on biomass fuel supply chains in Finland. And Upadhyay et al. (2012) used a 16  cost minimization approach of supply planning to determine the feasibility of situating a biomass power plant in Ontario. As indicated above, the use of optimization in the forest biofuel supply chain is often at the strategic level concerning the design of transportation networks. Once established these models are capable of ―what if‖ analysis, similar to simulation. Gauch and Gronalt (2011) for example used their optimization model of Austria’s forest biomass supply chain to assess the impact of rising energy costs on forest biomass procurement costs. Beaudoin, Frayret and LeBel (2008) used a cost minimization function of a biomass transportation system to identify infeasible plans at the tactical level, but acknowledged that the method was not appropriate for operational planning.  Typically when discussing the optimization of forest biomass supply chains researchers are identifying which supply points should be delivered to which demand points. While these can be large and complex networks such as Finland’s entire forest biomass supply chain (Ranta & Korpinen, 2011), they do not usually consider operational issues such as vehicle scheduling. However, through the implementation of dynamic programing, optimal volumes of delivered and stored material can be determined for given periods, as Shabani and Sowlati (2013) showed for a power plant in Canada. Niquidet, Stennes and van Kooten  (2012) used a similar approach to determine the costs of supplying a power plant in Quesnel, BC with forest harvest residues over a 25 year period and found that costs would increase considerably during the planning horizon. Sensitivity analysis of optimization models also has the power to reveal additional valuable information. Moisture Content for instance, has been shown to correlate with cost, where delivering dryer material can result in reduced costs (M. B. Alam, Pulkki, Shahi, & Upadhyay, 2012). Regardless of the methods employed, the process of developing a model 17  provides the researcher with great insight and understanding into the system they are analyzing (Gautier, Lamond, Paré, & Rouleau, 2000). The comparison of simulation and optimization techniques for studying biomass supply chains is not a new concept, with each method having been shown to have its own advantages (De Mol, Jogems, Beek, & Gigler, 1997). Combining simulation and optimization to study biomass supply chains is a more recent approach to analyzing these systems. By using the output of the simulation platforms FPInterface and Optitek as inputs into an optimization module called LogiOpt, researchers were able to identify opportunities for supply chain improvement that were overlooked by planners who focus on smaller pieces of a system (Morneau-Pereira, Arabi, Gaudreault, Nourelfath, & Ouhimmou, 2013). Alternatively, a series of optimization models can be run to identify strategic, then tactical, then operational optimums followed by simulating the long term implications of these decisions; as was done by Gautam et al. (2014) to show improved supply chain performance.  Agent based simulation methods have the ability to incorporate optimization. By allowing agents access to all available information regarding the supply chain they are part of, they can be programmed to make decisions which will contribute to a global optimum. Vahid (2011) used this methodology to investigate the value of information within BC’s forest industry supply chain and assess the impacts of changes to timber harvest policy on the supply chains performance.  While simulation and optimization have been discussed at length as tools for supply chain planning they are by no means the only tools available to analyze these systems. Gronalt and Rauch (2007) for instance, used a simple stepwise heuristic approach to identify a least cost 18  forest fuel supply network design. This included a comparison between on-site and centralized grinding, as well as the impact of adding intermediary terminals to the system.  Spatial analysis is also a common approach to determining forest biomass supply costs and availability. This involves using Geographic Information Systems (GIS) and parameters to relate identified areas to supply volumes. Costs are then applied based on assumed processing costs and transportation costs for each area. This approach is often a precursor to optimizing a supply network (M. B. Alam, Pulkki, & Shahi, 2012). Spatial analysis of this sort is a very commonly used tool in the field of forest and biomass energy planning and has been employed in Canada (Ralevic, Ryans, & Cormier, 2010), the U.S. (Perrin, Sesmero, Wamisho, & Bacha, 2012), Japan (Kinoshita, Inoue, Iwao, Kagemoto, & Yamagata, 2009), Spain (Panichelli & Gnansounou, 2008), and Portugal (Viana, Cohen, Lopes, & Aranha, 2010).  With growing concerns over the environmental impacts of goods and services the field of Life Cycle Analysis (LCA) has flourished. These methods aim to calculate the total embodied energy and emissions associated with a product beginning with all its raw materials through to its final disposal. The complexity of this type of analysis is highly dependent on the expansiveness of the supply chain in question. LCA has been used to study the costs and emissions associated with biomass energy from agricultural (Morey, Kaliyan, Tiffany, & Schmidt, 2010) as well as forest residues (Murphy, Devlin, & McDonnell, 2014). Another general approach to comparing alternatives within a supply chain is the economic Input/Output (I/O) model. These are total cost summations of a discrete number of alternative scenarios, often with the objective of identifying the least cost alternative. I/O models can also be referred to as engineering economic analysis, as in the case of determining the feasibility of a biomass power facility in Quesnel, BC (Kumar, Flynn, & Sokhansanj, 2005). 19  These methods are common in examining forestry and biomass supply chains. For BC’s forest biomass industry Lindroos and Sowlati (2011) compared roadside grinding, residue bundling and loose slash transportation by this method.  Related to several of these modeling and analysis methods is the concept of biomass supply curves. These often relate the delivered cost of material or the net cost of producing bioenergy to another metric of interest such as transportation distance or facility capacity. Kerstetter and Lyons (2001) developed forest biomass supply curves for the U.S. relating delivered cost to travel distance and regional operating factors such as slope. Supply curves can then be used to estimate the total economically viable biomass available in a region, such as the entire U.S. (Kocoloski, Michael Griffin, & Scott Matthews, 2011).  Models are valuable tools for analyzing supply chains and providing information to decision makers at all levels. As discussed above there are a wide variety of these tools, each well suited to perform a specific task or range of tasks. Simulation and optimization were discussed at length as they are arguably some of the more powerful methods of investigating supply chains, with optimization being well suited for network design and simulation particularly adept at operations level analysis. 1.6 Moisture Content When producing energy from woody biomass, moisture content is the most important quality factor. MC has been found to impact the storage properties and transportations costs of biomass logistics systems (Acuna, Anttila, Sikanen, Prinz, & Asikainen, 2012). Biomass energy systems are sensitive to the MC of the input material. When this material has a higher MC more energy used within the system, or in a pre-drying phase to reduce the amount of water in the 20  biomass and enable its energy release. This results in an inverse relationship between MC and heating value (Alakangas, 2005). In short, dryer wood burns better. Different biomass energy technologies have varying sensitivities to MC. With some systems, a MC outside of its optimal range can lead to a significant reduction in efficiency of energy production. Biomass boiler systems, where the material is directly combusted, can operate with a MC of up to 65%. However, the more homogenous the material, the better it will run. Other systems such as downdraft gasifiers require substantially dryer material, preferably below 20% (USEPA, 2007). Moisture content is commonly expressed in one of two ways; either green basis or ovendry basis. Green basis is the convention most often used when discussing woody biomass, whereas ovendry basis is the preferred method of calculation when discussing solid wood products. Throughout this work any discussion of MC is referring to green basis which is given by Equation 1 (Govett, Mace, & Bowe, 2010). Equation 1 Moisture Content of Wood     (           )   (                                             )       Forest biomass MC is a subject that has been studied for decades in regards to fire risk and behavior (Viney, 1991). Methods have been developed for estimating surface biomass MC using remote sensing technologies to assist with the mitigation and management of forest fires. This can be done by analyzing infrared and broad spectral information to relate surface temperature and vegetation type to water content and drought/ fire risk (Verbesselt, Fleck, Coppin, & Viegas, 2002). Prior to these advancements in remote sensing technology field sampling was carried out to assess fire risk. Standardized practices for determining moisture 21  content of forest materials have long been in place (Norum & Miller, 1984). These procedures for sampling, weighing and oven drying material are still in use today as the most accurate way to obtain biomass MC. In British Columbia’s coastal temperate climate limited research has been conducted on woody biomass MC. Traditionally harvest residue piles across the province were burnt on site to reduce wildfire risk and clear the areas for replanting. FPInnovations, Canada’s not-for-profit forest research organization, has undertaken several studies of harvest residue MC in the BC coastal region. Baxter (2008) tracked MC of harvest residue piles for the purposes of obtaining more complete combustion and reducing smoke during pile burning. Once interest began to grow in using these residues for bioenergy production the research began to focus on the available supply and the MC for energy purposes. This involved assessing road accessibility for the equipment required to process and transport hog fuel and determine a relationship between the amount of biomass removed from a cutblock in relation to the harvested timber volume, referred to as a biomass ratio (MacDonald, 2009). Dyson (2013) then conducted sampling of debris piles to investigate the drying trends and seasonal variation of MC in residue piles. While this study provided a rather limited snapshot of the MC of BC’s coastal residue piles, it provides valuable information in calibrating a seasonal MC schedule for BC’s coast. While other regions in Canada have benefited from increased sampling to clarify seasonal variations in MC (S. Gautam, Pulkki, Shahi, & Leitch, 2012), the resulting data is not applicable to BC’s coast due to significant climatic differences. The sampling methodologies employed by Gautam et al. could however be of use in conducting future studies on BC’s coast. In Europe where forest biomass has been considered as an energy resource for decades, the body of research on harvest residue MC is more extensive than that of North America.  Jirjis 22  (1995) studied the effects of storage on MC and dry matter loss. Both comminuted hog fuel and unprocessed logging residues were investigated, with various storage options for each. Typical forms of storage for unprocessed logging residues are windrows with or without covers, and bundles. Once the material is comminuted an anaerobic environment is created where dry matter losses are accelerated. Whereas with unprocessed logging residues, particularly when covered, dry matter losses are limited to below 1% per month.  Further research was conducted by Nurmi (1999) of Norway spruce logging residues in Finland. The MC of the material was monitored under different storage conditions for 1 year. It was determined that the residues are best left to dry on site to reduce the MC and limit dry matter losses. Similarly, Pettersson and Nordfjell (2007) tracked the MC of logging residues in Sweden under varying storage conditions including compaction of residues. They found that MC of residues can drop rapidly in good weather, also that the remoistening of residues during the winter occurred to a greater extent in loose residues rather than compacted.  With increasing demand for bioenergy feedstock the MC trials of logging residues under various storage conditions continued. Röser et al. (2011) ran a series of trials across Europe in Finland, Italy and Scotland. They investigated the changes in MC of roadside piles, bundles, partially debarked logs, and covered material. It was found that the effectiveness of each storage method on reducing the residue MC varied by species and region.  A culmination of decades of research on logging residue MC in Europe has generated some important findings. Sikanen (2010) for example, identified that transportation costs are greater for material with a higher MC due to the increased amount of water being transported. The development of a forecasting algorithm for drying of logging residues  (L Sikanen, Röser, Anttila, & Prinz, 2012) is however one of the most useful tools that’s been developed from this 23  collection of research. Although residue MC is affected by highly variable factors such as climate, storage type and time of timber harvest, this work shows a trend in MC behavior. With this trend and the data generated by Dyson (2013) a model for MC in BC’s coastal residue piles can be approximated.   24  2 Methods A simulation model was developed to study biomass grinding and transportation operations, while incorporating moisture content as a material quality factor. The scope of the simulation model is limited to sourcing forest harvest residues from a select set of cutblocks on the south west coast of British Columbia and delivering them to the Silverdale reload facility located on the Fraser River within the district of Mission, BC.  From Silverdale the residues are transported by barge to HSPP, however these secondary reloading and barging operations were not included in the scope of this analysis. 2.1 AROMA Simulation Model AROMA is a modeling platform developed using JaamSim discrete event simulation software which has been specifically tailored for natural resource supply chains (King & Harrison, 2013b). AROMA is an agent based system programed in Java code, enabling the development of simulation models using a high-level language customized for supply chain model development. Models are developed by generating a network and parameterizing objects to engage in specified behaviors within that network. Networks themselves can be quite complex with unique attributes specified to each segment. Once the data for all system attributes is collected it is transformed into AROMA code using Excel spreadsheets. These blocks of code are then copied into Notepad++ and assembled to generate the simulation model. The computations of the simulation model are displayed visually in the form of trucks and machines working on a map. This portion of the simulation model is programed alongside the development of the system attributes. By using 3D collada files for equipment and materials, machine behavior can be observed at the site level. A record is also generated of all the object states and behaviors in a spreadsheet format at a one second interval. 25  2.2 Simulation Output Processing and Analysis Tool In order to process and summarize the simulation model’s output spreadsheets, a visual basic program was created in MS Excel. This program scanned each of the model output spreadsheets and calculated the total time spent by each objects on all activities. Summary values were generated at the equipment level and system level. At the equipment level the following summary values were generated: number of cycles per truck, distance traveled per truck, time grinders and loaders spent walking between cutblocks, average cycle time per truck, average cycle distance per truck, and the total fuel consumption and emissions for each truck and piece of equipment. To complete further analysis the values calculated at the overall system level included: total time to complete, average delivered MC of material, total number of cycles, average number of cycles, average cycle time, total fuel consumption and emissions by trucks, grinders and loaders, number of operating days, average cycle time, total fuel consumption and emissions, working time by fleet, cost of trucking, grinding and loading, estimated mobilization costs, total delivered volume and tonnage, total overall costs, trucking, grinding, loading, mobilization and total costs per ODT, hours of labor only for trucking, grinding and loading, and hours of labor and equipment for trucking, grinding and loading.   26  3 Study Area Description 3.1 Climate and Forest Type The south coast of British Columbia is situated in a temperate rainforest characterized by high annual precipitation and steep mountainous areas. The case study area includes the mountains surrounding the Fraser Valley and Fraser Canyon, directly to the East and North East of Vancouver, BC. This region and the cutblock sites included in this investigation are predominantly contained within the Coastal Western Hemlock Biogeoclimatic zone. These forests are dominated by Western Hemlock (Tsuga heterophylla), Western Red Cedar (Thuja plicata), and Coastal Douglas Fir (Pseudotsuga manziesii). The study area also includes minor components of the Engelmann Subalpine Fir zone and the Mountain Hemlock zone. As a result of historic logging and development in the region much of the existing forest is second growth. Table 1 Biogeoclimatic Zones of Study Area BEC Zone Cutblock Area (ha) Proportion of Total Cutblock Area CWH 1865 95.0% ESSF 69 3.5% MH 30 1.5%  27   Figure 1 Map of Biogeoclimatic Zones in Study Area 3.2 Infrastructure and Forest Operations Industrial logging in the Fraser Valley began in the 19th century (Fraser Valley Guide, 2015) leaving a legacy of 2nd growth forests and an extensive network of access roads. Some of these roads are still in use today as highways and active resource roads, others have returned to 28  an undeveloped forested state. The continuation of timber harvesting operations has left a well-developed network of active resource roads. Due to the mountainous terrain, much of this network is characterized by mainline roads running along valley bottoms with spur roads switchbacking up the mountain sides to access higher elevation stands. Operations in the area range from ground based skidding and hoe chucking, to cable yarding, to helicopter yarding. In the case of ground based and cable operations it is generally standard practice in the region to skid or yard whole trees to the roadside prior for bucking and limbing. This results in large piles or windrows of harvest residues remaining on landings or along the side of in-block roads. These piles have typically been burned on site to reduce the future fire hazard and clear the area for replanting. 29   Figure 2 Map of Study Area's Road Infrastructure    30  4 Input Parameters 4.1 Geospatial Network The simulation model is composed of a series of supply points connected to a central demand point by a series of transportation routes. In order to develop the model, the locations and attributes of these points and lines needed to be determined. This involved identifying the locations of currently available harvest residues, determining the required access roads, and applying various net downs to produce a final network for this case study. Within the region there are several crown land forest licensees and land owners with timber harvest rights. However as forest licensees in British Columbia have legal obligations to manage their harvest residues, specific agreements often need to be in place to access these materials. As a result, this case study focusses on residues identified by BC Timber Sales (BCTS) and the Malcolm Knapp Research Forest (MKRF) with whom access had been negotiated by a biomass fuel supplier. In total 102 cutblocks were identified by BCTS and 13 by MKRF as having harvest residues on site. Once a list of cutblocks was established, geographic information was required for further analysis and to establish a network for the model. This geographic data for all BCTS cutblocks was obtained from DataBC, a provincial government website which hosts a variety of public information. This dataset includes dozens of attributes for each object including information on the forest license and cutting permit. Data for all MKRF cutblocks was retrieved from the research forest’s administration office. A geographic data set of all roads in the province was also retrieved from DataBC. The pertinent attributes associated with this data included road surface type and road class. This 31  dataset was trimmed to include only the roads required to connect the identified cutblocks for ease of management.  4.1.1 Road Accessibility Upon assembling the geographic data on all potential cutblocks and roads the information was imported into ESRI’s ArcMap 10.1 to be edited. Using ArcMap, kml files were generated and imported into Google Earth. This allowed for roads and cutblocks to be inspected on a three dimensional landscape with aerial imagery in the background. By viewing the data in this way details such as slope and harvest method for individual cutblocks and road segments could be more easily interpreted. This information was combined with a practical understanding of forest road and cutblock layout, and operational harvest planning to generate a number of assumptions about each specific cutblock. Forest roads are typically designed to accommodate the specific vehicles and equipment required for timber harvesting rather than the less versatile residue transport vehicles.  As a result many cutblocks are unavailable for residue collection, particularly in BC’s mountainous coastal region. FPInnovations has studied the accessibility of cutblocks for residue transport vehicles on Vancouver Island and BC’s South coast. In assessing the accessible biomass for the Campbell River region of BC, they ―estimated that 60% of the areas suitable for ground-based harvest systems and 30% of the areas suitable for cable harvest systems were accessible by chip vans.‖ (Macdonald, 2012) In order to apply this accessibility net down an estimate of the likely harvest method had to be made for each cutblock. Using Google Earth and the imported geographic data, cutblocks were assessed as using either ground-based, cable or helicopter harvesting system. When evidence appeared that both ground-based and cable harvesting systems were used, it was 32  considered a cable harvest cutblock. These assessments were made by analyzing the ground profile, considering in block road patterns, inspecting apparent harvesting systems of neighboring cut blocks, and in some cases areal imagery was current enough to inspect the cutblock directly. While these methods may not be 100% accurate they were deemed sufficient to develop this case study and estimate the accessible volume of harvest residue.  Further to the estimation of the harvesting system, other notes were made on a cut blocks accessibility. Some blocks were deemed less likely to be accessible due to a series of back to back switchbacks in their access road. Others required barging across Stave Lake, BC for access. This would add significant costs to the collection of harvest residues and were deemed unlikely to be targeted. Each cutblock was then ranked as being Probably Accessible, Maybe Accessible, or Probably Not Accessible. Of the original list of 115 cutblocks, 8 required barge access and 4 used helicopter harvesting systems. These were removed from the list of supply blocks due to their operational constraints. From the remaining 62 ground-based cutblocks, 40% were removed to meet the accessibility estimates used by MacDonald et al. And of the 48 cable-based cutblocks, 70% were removed.  These removals took place in 2 stages. First those ranked Probably Not Accessible were removed, then those ranked Maybe Accessible. In total 24 ground-based blocks and 20 cable-based cutblocks were removed in this manner. Of the remaining cutblocks 1 ground based block was randomly removed to meet the 40% reduction, and 14 cable-based blocks were randomly removed to meet the 70% reduction. This left a total of 51 cutblocks accessible for harvest residue collection, 37 of which were ground-based and 14 were cable-based. Preparation of road network data began using ESRI’s ArcMap 10.1. The road network data obtained from DataBC (BC Government, 2013) was trimmed down to the shortest direct 33  routes between each cutblock and Silverdale. Among the attributes of the DataBC road network were ROAD_CLASS and ROAD_SURFACE. These attributes were used to help assign 1 of 3 speed classes to each road segment; Gravel, Paved or Freeway. Discussions with industry experts helped identify average trucking speeds for each of these road types given the local terrain and truck configurations currently employed (McIver, 2013b).  Table 2 Simulation Model Network and Object Attributes Network Road Points Network Road Segments Cutblocks Reported Objects Vertex ID Road ID Cutblock ID Trucks Road ID Road Name # Stockpiles in Block Containers Road Name Speed Class m3 Residue per Stockpile Grinders Speed Class Road Length X Coordinate Loaders X Coordinate   Y Coordinate Unloaders Y Coordinate     Stockpiles  The spatial data was then converted into a series of road segments broken up by nodes at junctions and changes in Speed Class; connecting all cutblocks with Silverdale. Each road segment was then broken down further into a series of X, Y vertices in order to be input into the simulation model.  The resulting data was converted into 2 blocks of code in the same way as the cutblocks, one for road segments and one for road vertices. Table 3 Road Network Speed Classes Speed Class Average Speed Gravel  15 km/h Paved  70 km/h Freeway 90 km/h  In order to run the simulation model the road segment data was linked together in series. Additional code was generated to identify which road segments were directly connected to which nodes. This allowed for vehicles and equipment to travel across the road network in a fluid 34  manner. Two-way travel was then enabled on paved and freeway road segments, while frequent pullouts were added to the single lane gravel roads. Trucks and equipment were then able to move past each other in opposite directions on the two way roads while having to wait at pullouts for others to pass on gravel roads. In addition, gravel roads were delineated between those which a machine would be able to travel along on a lowbed and those on which it would walk. This was done by analyzing the spatial data in ArcMap and Google Earth to identify neighboring cutblocks which machinery would likely walk between rather than incur the additional expense of hiring a lowbed. 4.1.2 Supply Attributes With the supply point cutblocks identified, the availability of harvest residues needed to be determined for each location.  The first step was to collect all available data on the cutblock areas. This information differed between the BCTS cutblocks and the MKRF cutblocks.   Information on the BCTS blocks was procured from the online public databases DataBC (BC Government, 2013) and Harvest Billing System (HBS) (BC Ministry of Forests, 2013).  The geospatial data contained in the BC Forest Tenure Authority GIS layer included cutblock area, total cutting permit area, and additional forest licensing information. By entering a cutblocks permit information into HBS, the timber cruise volume and species mix for the full cutting permit was obtained. To calculating an average volume, in m3 per hectare, each cutblocks volume and species composition was estimated. This method assumes a relatively homogenous timber type across each cutting permit. While this may result in a slight deviation from actual cutblock volumes, the total estimated volume for a region should relatively accurate. For the MKRF cutblocks the timber harvest information including cutblock volume and species, was obtained from the research forests administration rather than BC’s public databases. 35  Biomass ratios were used to calculate a cutblocks forest harvest residues available for collection. These were developed by FPInnovations for similar climactic regions on nearby Vancouver Island (Macdonald, 2012). These ratios estimate the available timber harvest residues located at roadside as a proportion of the merchantable timber harvest. The ratios are dependent on species, stand type (second growth or old growth), harvest method and log processing method. Table 4 Biomass Ratios (Macdonald, 2012) Stand Attributes and log-handling methods Biomass ratio % Ground and cable harvesting Aerial harvesting Logs to roadside Trees to Roadside Logs to roadside Trees to Roadside Hemlock Old Growth 13 15 2 2   Second Growth 10 15 2 2 Cedar Old Growth 8 10 2 2   Second Growth 9 15 2 2 Fir Old Growth 10 12 2 2   Second Growth 9 15 2 2   Discussions with industry experts indicated that the identified cutblocks were all likely to be second growth (McIver, 2013a). As well all the identified cutblocks were likely to involve roadside bucking (Lyons, 2013). In order to estimate the mass of harvested timber the oven dried weight of each primary species was obtained from the USFS (Miles & Smith, 2009).  Table 5 Average Oven Dried Weight of Local Wood Species Species Green Volume Basis (kg/m3) Cedar 310 Fir 450 Hemlock 420  36  Given each cutblocks primary species, merchantable timber volume, assumed stand type and bucking configuration, biomass ratio, and oven dried density, the volume of collectable biomass was then calculated.  Figure 3 Map of Road Network  37   For the transportation of biomass it is important to know both the volume and weight of the material. A cutblocks residue volume will be constant but as the material moisture content changes the resulting material weight with vary. In the process of grinding the residue from the form of branches, tree tops and stumps into hog fuel the density of the material is significantly altered. The average oven dried hog fuel density of 194 kg/m3 was used for this processed material. This was calculated given the conversion of approximately 1.1 Oven Dried Tonnes (ODMT) being equal to 1 Volumetric Unit (VU) (McIver, 2013c) and 1 VU equaling 5.66m3 (Briggs, 1994).  4.2 Equipment & Costs Equipment Specifications and production costs were sourced from industry experts (McIver, 2013a), a site visit to an active operation, and relevant literature (MacDonald, 2009), then applied in the form of hourly rates which included capital and operating costs (Table 6).  These costs were broken into active costs which were represented by the full rate, and idle costs which assume all labor and capital costs but no fuel consumption.  Equipment types included grinders, loaders, unloaders and trucks. These objects were then assigned rates. In order to assess the impact of machine productivity on biomass grinding operations two types of equipment were considered, highly productive (HP) more costly equipment and lower productivity (LP) less costly equipment. HP grinders and loaders which operate in tandem were set to operate at 169.9 m3/h, and unloaders at 372 m3/h. While LP grinders and loaders operated at 87.74 m3/h and unloaders at 372 m3/h. Trucks speeds were programed to vary based on the road type they are operating on, and their capacities were limited to 26.5 tonnes and 93 m3 for HP and 14 tonnes and 49.13 m3 for LP. The values for HP equipment were determined through discussion with industry experts (McIver, 38  2013c) and by observing operations. Whereas values for LP equipment were sourced from (MacDonald, 2009). Table 6 Equipment Hourly Costs and Capacities With and Without Fuel Consumption Equipment Idle Cost Active Cost Capacity High Productivity Diesel Configuration   Grinder $257.47 $341.47 169.90 m3/h   Loader $102.99 $136.59 169.90 m3/h   Truck/Lowbed $85.40 $125.00 26.5 tonnes / 93 m3 Low Productivity Diesel Configuration   Grinder $126.77 $210.77 87.74 m3/h   Loader $94.99 $128.59 87.74 m3/h   Truck/Lowbed $70.40 $110.00 14 tonnes / 49.13  m3 Centralized Electric Grinder Configuration   Grinder ø $150.00 169.90 m3/h   Loader $102.99 $136.59 169.90 m3/h   Truck/Lowbed $85.40 $125.00 26.5 tonnes / 93 m3 Common Parameters   Highway Diesel   $1.32/L      Equipment Diesel   $1.05/L     Electricity   $49.90/MWh     Unloading Rate     372 m3/h  System costs were determined by defining model object states as having either no cost, idle cost, or full cost. When an object with an assign rates enters a full or idle cost state it incurs cost at its assigned rate. Operations were assumed to continue 24 hours a day as per current practices of observed operations, with the loaders and grinders nearly always in use. Their active cost was calculated by the summing the working time and the time spent positioning for the truck. The loader and grinders idle time occurred only when they were sitting idle and there were no trucks queued.  In order to investigate a centralized grinding approach additional equipment was included in the model. Both the low and high productivity equipment had a loader placing logging debris 39  into horizontal grinder, which then throws the comminuted biomass into a waiting truck. Alternatively, the loader could place logging residue directly into a truck to then be delivered to a central location for grinding. To model this it was assumed that the same trucks could be utilized for the transport of logging residue but a lower material density of 144.9 kg/m3 was used (Cross, Turnblom, & Ettl, 2013). The new configuration assumed the use of an existing electric grinder at the industry partners sawmill site adjacent to the Silverdale reload facility. This required an additional loader at Silverdale to load the logging residues into the electric grinder, while eliminating the need for the diesel grinder at the cutblock. As the electric grinder performed the existing function of processing sawmill offcuts at less than 100% capacity it was assumed that its costs were only attributed to the logging residue processing operation while it was active on that task. The loader servicing the electric grinder was split into active and idle costs as with the loader servicing trucks at the cutblocks.  Trucks began incurring costs from when they were initially dispatched until their final return to Silverdale. Their idle time was calculated by summing the time spent parked, being scaled, queued, loading and unloading. Their active time was calculated by summing their time spent traveling, and pre and post loading. Equipment relocation costs were broken into two categories; walking and lowbeding. This required a subjective estimate of when operators would choose to walk their machines between neighboring cutblocks and when a lowbed would be brought in to relocate the equipment. In general it was assumed that machines would walk between cutblocks within a block group area. However, each block group was analyzed individually and if a machine had to 40  be relocated along a section of paved road, or if the distance was greater than 3km, in most cases a lowbed would be employed in addition to the equipment idle costs. All operating costs were then summed to generate a total cost for grinding, loading and transporting the biomass in this case study. The total cost was then divided by the total delivered oven dried tonnes of biomass to give the cost in $/ODT. This metric was then used to identify the least cost configuration and perform additional sensitivity analysis.  4.3 System Behavior: Schedules and Delays Schedules are generated to direct the activities of all equipment. These include daily and annual operating periods, the order in which cutblocks are processed, and the predicted moisture content of harvest residues over time. Once the spatial network, objects, delays and schedules are programed, the simulation model could be run.  A model run begins at the start of year 1. At the time specified by the operating schedule the processing equipment is initiated at Silverdale and dispatched to the first assigned cutblock. As operations begin, a set number of trucks are dispatched from Silverdale at a specified Dispatch Delay of 40 minutes and continue to service their assigned equipment set while the operating schedule is active. Once the equipment processes and loads all the material in all stockpiles at a cutblock it relocates, either by lowbed or walking to the next assigned cutblock. These processes continued until all the cutblocks had been processed or the operating schedule dictates a halt to operations. A visual representation of the simulation model’s calculations can be viewed while running at any speed, including real time. The model results are output as a series of individual spreadsheets, one for each of the objects within the simulation; which ranged from 14 to 64. These outputs depict the activity of every object in the system at a 1 second time step. This data 41  is then processed and summarized to complete a sensitivity analysis of various system components. The Simulation Output Processing and Analysis Tool enabled the model to be calibrated to better reflect actual residue grinding operations. This required the development refining of several schedules and model commands.  A list of destination cutblocks was assigned to each set of grinders and loaders being dispatched. For initial model runs this was simply an alphabetical list of cutblocks being processed by a single loader and grinder. As calibration proceeded, intuitive logic was applied so equipment movements could be minimized. Additional grinders and loaders were then added and their scheduled cutblocks were adjusted until each equipment set covered its own geographic region and the time taken to complete their assigned cutblocks was about equivalent between equipment sets. This was done by iteratively adjusting which cutblocks were processed by each equipment set until each fleets working time was as close to the others as possible, and the overall time to completion was at its minimum. Along with a cutblock list for each fleet, a number of schedules and values were assigned to the system as a whole. These included an operating window, delays and queues, and a MC schedule. The operating window was set to begin operations June 1 at 6 am. This coincided with the scheduled start of real world operations. From that point operations were set to continue at 24 hours per day until the end of October when it was anticipated rain and snow would become operationally prohibitive. This schedule considers running 2 shifts with backup equipment to be utilized while regular maintenance is performed in order to maintain 24 hour operations. This configuration was taken into account when developing the hourly costs of operations. With an operating window in place, model behaviors such as delays needed to be programed for the system to reflect operational realities. This then allowed objects within the 42  model to queue so issues with traffic congestion could be examined.  At Silverdale a scaling delay of 10 minutes was determined to be reasonable by monitoring actual trucking activity. This paused the trucks at Silverdale for 10 minutes before they could begin unloading. At the cutblocks a 5 minute delay was added before and after loading to capture time lost while positioning the truck and grinder, and preparing the truck for highway travel once loaded. These were again identified by monitoring actual trucking operations. Additionally, a delay of 8 minutes was added at the cutblocks for the grinders and loader to relocate themselves within the cutblock to their next stockpile of biomass.  In order simulate delays which occur during biomass processing and loading operations the volume of material on each cutblock was divided between stockpiles. A Travel Time Delay was then assigned to account for the time equipment must take to relocate between stockpiles.  The number of stockpiles on a cutblock block was estimated based on its area. A sample of 10% or 13 cutblocks was analyzed for their shape and the configuration of their internal road network. These blocks consisted of between 1 and 5 distinct roaded areas between which a machine would have to spend non-productive time relocating.  On average these sampled blocks contained 0.154 areas per hectare. This figure was applied to all the cutblocks and rounded to the nearest whole number to approximate the number of stockpiles in each block, ranging from 1 to 6. Discussion with industry experts (McIver, 2013c) then identified an approximate travel time of 5 to 10 minutes between the block areas. As a result an average travel time of 8 minutes was assumed for equipment to move between stockpiles within a cutblock. This then formed the equipment Travel Time Delay within the simulation model. As with the cutblocks, the demand point at Silverdale was assigned a stockpile. However the Silverdale stockpile was given an initial volume of zero. Internal road segments were 43  generated for Silverdale by tracing over aerial imagery from Google Earth, along with que points for truck scaling and unloading. In the simulation model trucks are then able to enter the Silverdale reload, pause on the scale for a set time, then proceed to unload at an appropriate location before departing the facility. Silverdale’s roads were given a speed limit of 6 km/h, while truck unloading occurred at a rate of 372 m3/h. And a Post Loading Delay of 5 minutes was assigned to account for trucks reconfiguring for highway travel after unloading is complete. Delay times and operating rates at Silverdale were determined by observing operations and consultation with industry experts (McIver, 2013a).  For trucks be dispatched to a cutblock they needed to be assigned to each equipment set and given a dispatching interval. This Dispatch Delay between trucks departing from Silverdale was determined iteratively to minimize the time trucks would spend queued at a cutblock waiting for their turn to be loaded. After trying 60, 6, 30, and 45 minutes, 40 minutes was found to reflected desired operational behavior of a truck arriving within a few minutes of the previous one completing the loading process. This value is also quite close to the average loading and delay time for a truck of roughly 39 minutes. When the Dispatch Delay was shortened trucks experienced excessive queuing at the cutblocks as they wait to be loaded. And when the Dispatch Delay was lengthened the grinder and loader experienced excessive downtime, resulting in an inefficient use of equipment. The final schedule applied to the system was the monthly average moisture content (Figure 4). When a load of biomass was processed and transported to Silverdale it was assigned the moisture content of the current month. The moisture content of the load was then fixed for the duration of the simulation. Once the material was added to the stockpile at Silverdale it was 44  blended with all the biomass that had already been delivered and would impact the average moisture content of the delivered stockpile. 4.4 Moisture Content While there have been several case studies of harvest residue’s moisture content in Canada, the topic has been explored much more thoroughly in Europe. Not only has field sampling been conducted across the continent (Röser et al., 2011), but models have been generated to predict the moisture content of various types of forest harvest residues with the expressed intention of helping plan biomass collection operations (Acuna et al., 2012). The moisture content schedules developed in these studies varied based on tree species and region. They were compared with the average data points found by Dyson for coastal BC and the best fit curve from Acuna was selected to be used as the moisture content schedule for coastal BC. Table 7 Harvest Residue Moisture Content Sampling Results on Vancouver Island (Dyson, 2013) Location MC by Sample Period June August January Nanaimo 37.25% 23.75% 36.50% Campbell River 37.25% 25.50% 36.50% Port McNeill 44.25% 34.25% 44.25% Port Alberni 48.75% 37.00% 50.00% Average 41.88% 30.13% 41.81%   45   Figure 4 Anticipated Seasonal Moisture Content Variation of Harvest Residues on BC's Coast (Acuna et al., 2012; Dyson, 2013)   46  5 Results  5.1 Fixed Fleet Planning The first step in determine the least cost configuration of delivering biomass to Silverdale was to simulate a range in the number of trucks and simultaneous operations. This began with a single loader and grinder equipment set, with trucks added one at a time. Once the costs per tonne began showing a clear and sustained increase with each additional truck the system was reset to a minimal number of trucks servicing two equipment sets. The iterative process was then repeated adding one truck at a time to the operation taking the longest to finish. This resulted in alternating between the two active operations. Again when a clearly increasing cost trend was identified the number of trucks was reset and a third equipment set was added. The number of trucks in the system was then increased one at a time with each fleet gaining a truck every third model run. As the number of trucks servicing an operation was constant throughout a model run, this analysis was referred to as fixed fleet. Further analysis was conducted where the number of trucks was able to vary between cutblock groups, referred to as variable fleet. This fixed fleet analysis identified a least cost number of trucks for each equipment set configuration. With one operation, 6 trucks were able to deliver all the biomass for $72.08/ODT. In the case of 2 operations the minimum delivered cost identified was $72.90/ODT and with 3 operations it was $72.50. While these represent least cost fleet sizes for the given operations, it may be desirable to increase the number of trucks in use if time constraints demand so. In running simulations of each plausible trucking configuration expected cost curves were generated (Figure 5). With additional operations the curves displayed a certain level of non-linearity. This is due to the varying marginal benefit of adding trucks to certain operations. As 47  seen with 2 operations, adding a tenth truck has only a small impact on costs while adding an eleventh truck results in a noticeably greater cost reduction.   Figure 5 Delivered Cost of Biomass with High Productivity Diesel Equipment Once a minimum cost is reached for an equipment set, the addition of trucks begin causing bottlenecks in the system. Within the simulations visual interface it becomes apparent that trucks start backing up at the cutblock waiting to be loaded. With an increased number of vehicles, trucks also spend more time waiting in pullouts along gravel roads. This results in excessive queuing and an increase in transportation costs. As the number of trucks is increased towards the least cost position on the curve the benefits of increasing utilization of the grinding and loading operation exceed the cost of adding additional trucks. Beyond a point the diminishing value of increased utilization is outweighed by the cost of additional trucks (Figure 6). As a result of the additional traffic experienced with 2 and 3 operations, trucks occasionally 48  become queued at Silverdale waiting to unload. These factors result in the minor increase in minimum cost experienced in the multi-operation simulations.  Figure 6 Cost Breakdown of Biomass Delivery for HP Equipment and 1 Operation 5.2 Alternate Equipment  The type of equipment used in forest operations impacts the productivity and costs of those operations. The initial model calibration was completed using the HP equipment specifications reported by an active biomass processing operator. Alternatively, FPInnovations has conducted analysis on the processing and delivery of logging residue on the BC coast using the LP equipment configuration (MacDonald, 2009). While the equipment included in the FPInnovations analysis had lower loading and grinding rates, and a lower hauling capacity, it also carried lower hourly costs. In order to test the sensitivity of the case study system to the type of equipment used, the productivities and hourly rates from the FPInnovations study were substituted. 49  Once the equipment productivity and costs were included, the simulation was analyzed to determine if recalibration was necessary. Although the rate of grinding and loading was decreased from the initial assumptions, the smaller capacity trucks resulted in approximately the same average loading time of 40 minutes rather than 39. As a result the trucking dispatch delay of 40 minutes was deemed appropriate for the LP equipment. Once the model was prepared with the new parameters the simulation was run with both the fixed fleet and variable fleet configurations. With the fixed fleet size of 6 trucks and one operation it was found that the LP equipment required an additional 44 days to complete operations. The average cycle time only differed by approximately 4 minutes, but the average delivered cost of biomass was $26.08/ODT more expensive with the LP equipment at $98.16/ODT. Upon closer examination of the model results it was identified that the average queue time for a truck was 17.8 hours over the course of operations with the HP equipment, but 46.1 hours with the LP equipment. While the loading times were approximately the same, the dispatch delay allowed for trucks to arrive slightly before the previous one was completely loaded. Over the course of operations at a cutblock this slowly resulted in a backlog, increasing with each new trucks arrival. With the LP equipment a total of 2,775 cycles were required to deliver all the material, nearly twice as many loads as the 1,521 required with the HP equipment. This extension of operations enabled a backlog of trucks to build up to a significant operational factor resulting in increased costs. However, with the HP configuration the total cost attributed to trucks queuing was $9,142.56, or approximately $0.36/ODT. In the case of the LP equipment, while significantly greater at $19,464.76, this only amounted to $0.78/ODT. The remaining cost difference between the two categories of equipment is largely attributed to the significantly greater number of machine and truck hours required to 50  process and transport the biomass. The most significant differences between the equipment types were the number of operating days and the cost, with the smaller equipment taking nearly twice as long to complete operations (Table 8). Table 8 Least Cost Fleet Size and Costs for Fixed Fleet Configurations High Productivity Equipment Low Productivity Equipment Least Cost Fleet Size Delivered Cost ($/ODT) Total Operating Days Least Cost Fleet Size Delivered Cost ($/ODT) Total Operating Days 6 $72.08 64 6 $98.16 108 12 $72.90 33 12 $101.02 57 19 $72.50 21 17 $101.13 41   Figure 7 Delivered Cost of Biomass with Low Productivity and High Productivity Diesel Equipment 51  5.3 Variable Fleet Cutblock Grouping A limitation of the fixed fleet analysis conducted above is the reliance on a constant number of vehicles services an equipment set. This approach, which has found a least cost solution of 6 trucks reporting to 1 equipment set, has the same number of trucks traveling from Silverdale to both the nearest and furthest cutblocks. While on average this generates a least cost solution, it also results in excessive queuing of trucks at the cutblocks with the shortest travel time and excessive downtime for the loader and grinder at the furthest cutblocks.  To refine the system and enable the simulation of a varying number of trucks to service an equipment set, the cutblocks were grouped into sub-regional clusters, approximately equivalent to a forest license cutting permit (Figure 8). These block groups were then run in the model separately using the same iterative approach used to determine the least cost fleet configuration for the whole system. Once a least cost number of trucks was determined for each block group (Figure 9) the total processing, transporting and relocation costs for each block group were summed. This resulted in the simulated least cost operations for the system as a whole if the fleet size were variable from one block group to the next. The total delivered cost here was determined to be $68.64/ODT, $3.44 less than the $72.08/ODT found with a fixed fleet size (Table 9). It was also identified that the delivered cost (Figure 10) and least cost fleet size (Figure 11) were dependent on the average cycle time required to service the block group. The distinction between travel distance and cycle time is subtle but important for this case study as some block groups located close to Silverdale require a majority of their travel on slower gravel roads while some more distance areas are almost directly accessible by paved road or freeway. While average cycle time explains much of the 52  variability in least cost fleet size, the available block group volume and number of internal relocations are also expected to impact the delivered cost and number of trucks.  Figure 8 Map of Cutblock Groups 53   Figure 9 Least Cost Fleet Size Charts by Block Group  54  Table 9 Delivered Cost of Biomass for Different Equipment Types by Block Group Cutblock Group HP-Delivered Cost ($/ODT) LP-Delivered Cost ($/ODT) Cost Difference ($/ODT) Average Cycle Time (hours) AL $74.75 $100.39 $25.64 4.9 CB $77.97 $111.23 $33.26 5.8 CO $48.45 $66.75 $18.30 2.8 CQ $59.59 $85.12 $25.53 4.8 EM $59.71 $85.15 $25.44 4.3 HU $67.57 $93.60 $26.03 4.4 HW $57.18 $77.72 $20.53 3.4 JN $54.78 $81.25 $26.47 4.7 LO $62.37 $76.91 $14.53 2.6 MKRF $58.53 $77.65 $19.12 3.2 NK $103.99 $151.91 $47.92 9.9 NL $80.60 $97.69 $17.10 3.7 PE $65.45 $83.97 $18.52 4.5 RU $54.41 $81.65 $27.24 4.7 SK $76.86 $114.47 $37.62 7.1 SP $103.44 $148.75 $45.31 8.6 ST $96.14 $140.15 $44.01 7.9 WB $60.78 $85.17 $24.39 4.2 Weighted Average $68.64 $96.95 $28.25 5.1  55   Figure 10 Impact of Block Group Cycle Time on Delivered Cost   Figure 11 Impact of Block Group Cycle Time on Least Cost Fleet Size for HP Equipment When comparing the results of the variable fleet size model runs of the two equipment types, the least cost configuration often had the same number of trucks servicing a block group. As well, the average cycle time was found to be similar, though the HP equipment was found to 56  consistently take several minutes longer.  By comparing the difference in costs between HP and LP equipment at a block group level a trend was identified in relation to cycle time (Figure 12). This indicated that the cost penalty associated with the LP equipment is greater for more remote operations.   Figure 12 Cost Impact of Equipment Productivity and Cycle Time by Block Group 5.4 Centralized Electric Grinding The same process of modeling the range of fixed and variable fleet sizes was conducted using the CE grinding costs and productivities. With a fixed number of trucks servicing all the cutblocks the least cost solution was found to be 17 trucks servicing 3 simultaneous loading operations (Figure 13). In contrast to the in block grinding operations, transportation of the lower density material required an average of 1869 truckloads of logging residue rather than 1544 of comminuted hog fuel. While the total costs are similar between the HP and CE equipment 57  configurations, the absence of the on-site diesel grinders and increased number of truckloads results in transportation attributing a greater proportion to the delivered cost (Table 10). This is also indicated by a steeper cost curve associated with CE compared to HP (Figure 14).   Figure 13 Delivered Cost of Biomass with Centralized Grinding    58  Table 10 Delivered Cost Breakdown of Operations Under Various Equipment Configurations   High Productivity Diesel (HP) Low Productivity Diesel (LP) Electric Centralized (EC) Fixed Fleet Variable Fleet Fixed Fleet Variable Fleet Fixed Fleet Variable Fleet Operations 1 1 1 1 3 3 Trucks per Operation 6 4 to 12 6 4 to 13 17 (5,6) 3 to 8 Operating Days 64 60 108 101 27 26 Transport Cost ($/ODT) $40.83 $40.11 $61.15 $62.92 $47.52 $46.51 Grinding Cost ($/ODT) $28.60 $26.28 $34.98 $32.24 $20.69 $20.97 Mobilization Costs ($/ODT) $2.65 $2.25 $2.03 $1.79 $0.61 $1.25 Total Costs ($/ODT) $72.08 $68.64 $98.16 $96.95 $68.82 $68.73   Figure 14 Delivered Cost of Biomass Comparison between Electric Centralized Grinding and High Productivity Diesel Comparing the CE and HP equipment configurations by block group reveals significant variability in the cost difference (Table 11). In some cases the CE equipment is more costly, 59  while for other block groups the HP equipment carries higher costs. By comparing the cost difference between equipment types to the average cycle time of the block groups a weak trend is established (Figure 15). While the cost difference is likely dependent on other factors such as block group volume, it can be identified that block groups with shorter cycle times are generally less costly using the CE equipment configuration. In particular, the CE configuration delivers biomass at a lower cost for all block groups with a cycletime of less than 4.5 hours within this system. Table 11 Delivered Cost by Block Group Under Various Equipment Configurations Cutblock Group HP-Delivered Cost ($/ODT) CE-Delivered Cost ($/ODT) Cost Difference ($/ODT) Average Cycle Time (hours) AL $74.75 $69.18 $5.57 5.0 CB $77.97 $75.54 $2.43 5.9 CO $48.45 $45.79 $2.66 2.9 CQ $59.59 $68.18 -$8.59 4.9 EM $59.71 $59.54 $0.17 4.4 HU $67.57 $64.61 $2.97 4.5 HW $57.18 $53.17 $4.01 3.5 JN $54.78 $64.55 -$9.77 4.9 LO $62.37 $50.09 $12.28 2.7 MKRF $58.53 $52.42 $6.11 3.3 NK $103.99 $116.51 -$12.52 9.9 NL $80.60 $67.31 $13.28 3.8 PE $65.45 $68.55 -$3.10 4.6 RU $54.41 $64.94 -$10.53 4.9 SK $76.86 $89.38 -$12.52 7.2 SP $103.44 $102.70 $0.74 8.7 ST $96.14 $97.10 -$0.96 8.0 WB $60.78 $59.98 $0.80 4.3 Weighted Average $68.64 $68.73 $0.15 5.2  60   Figure 15 Cost Difference Between HP Diesel and CE Configurations by Block Group  5.5 Sensitivity Analysis After examining the impact of fleet and equipment configurations on the delivered cost of biomass a sensitivity analysis was conducted to investigate the impacts of the biomass volume assumptions and to further understand moisture content’s impact on the supply chain. These analyses were conducted on the least cost configuration identified for the HP equipment configuration of 6 trucks servicing one operation. 5.5.1 Biomass Volume  The assumed biomass ratios and their relevant inputs formed a significant portion of the model parameters. Some of the results identified above are also likely to be partially dependent on the available biomass volume. If these assumptions around harvesting operations, stand type, 61  and their geographical appropriateness are inconsistent with the actual volume of delivered biomass, the resulting system would vary from the modeled results. In order to assess these potential impacts a sensitivity analysis was performed on the available volume of biomass at each cutblock. Using the identified HP equipment least cost fleet configuration of 1 operation and 6 trucks, the available volume was increased and decreased by 10 and 20 percent for each cutblock. This equates to testing biomass ratios of 12, 13.5, 16.5 and 18 around the baseline of 15 which all cutblocks in the case study shared. It was found that as the available biomass at the cutblocks increased, the cost of processing and delivering that biomass decreased. For this case study, a 10% change in the estimated biomass volume could result in greater than a $1/ODT change in the delivered cost (Table 12). This is comparable to the impact of adding or removing a truck from the least cost fleet (Figure 16).  Table 12 Delivered Cost Sensitivity to Biomass Volume Volume Adjustment Equivalent Biomass Ratio Delivered Cost ($/OCT) -20% 12 $74.66 -10% 13.5 $72.97   15 $72.08 +10% 16.5 $70.54 +20% 18 $69.96  62   Figure 16 Cost Sensitivity to Biomass Volume Relative to Fleet Size 5.5.2 Delivered Moisture Content One of the primary objectives of this research was to study the impacts of moisture content on forest biomass collection operations. The focus of this was to understand how seasonality would impact the final delivered moisture content of the processed biomass. Previous research indicated the material moisture content varies throughout the year, and likely decreases into a second season in the forest. To test the sensitivity of the initial moisture content schedule, 3 approaches were used: collecting material in the second year, testing a series of constant moisture contents, and varying the moisture content schedule up and down 4.5%, or half the difference between year 1 and year 2 (Figure 17). A total of 8 new moisture content schedules were tested. 63   Figure 17 Moisture Content Input Schedules  The second year collection was simulated by shifting the original moisture content schedule forward by one year. Constant moisture contents of 20%, 30%, 40%, and 50% were all tested to simulate a more rudimentary approach to analyzing the case study system. And the vertically shifted schedules, which also include the second year collection, were generated to determine the sensitivity of expected delivered moisture content to the original input schedule. As with the analysis of volume sensitivity, the additional moisture content schedules were tested on the HP fleet configuration of 1 operation and 6 trucks. It was identified that as the MC of material increases, trucks begin to hit their weight limit before filling to their volumetric limit. For the trucks used in this case study with a volumetric limit of 93m3 and a weight limit of 26,500kg, this threshold is reached at a moisture 64  content of 31.5%. This phenomenon results in an increase in the number of required truckloads (Table 13) to deliver the same amount of material. This in turn leads to a longer operating period (Figure 19) and increased costs (Figure 18). Table 13 Delivered Cost and Operating Days In Response to Various Moisture Content Schedules Moisture Content Input Schedule Average Delivered Moisture Content Total Number of Cycles Operating Days Total Cost ($/ODT) Constant 20% 20% 1394 60.5 $67.95 Constant 30% 30% 1394 60.5 $67.95 Constant 40% 40% 1585 66.6 $74.77 Constant 50% 50% 1894 77.6 $87.11 Reduced  24% 1394 60.5 $67.95 Second Year Post-Harvest 28% 1394 60.5 $67.95 Midpoint  33% 1422 60.9 $68.60 First Year Post-Harvest 38% 1521 64.1 $72.08 Elevated 42% 1640 68.2 $76.69   Figure 18 Cost of Delivered Biomass at Various Average Moisture Contents 65   Figure 19 Operating Days Required to Deliver Biomass at Various Average Moisture Contents  5.6 Comparison of Methods In contrast to the simulation model, the High Productivity equipment set of inputs and assumptions were used to estimate the total cost of operations by means of an Input/Output model in Microsoft Excel. The processing and traveling rates, along with the travel times to each cutblock, and the available volume were used to calculate the required time to complete operations. This included the same operational delays to reposition, scale and unload trucks, as well as relocate equipment. The same hourly costs were then applied to each equipment types required working time. The resulting hours and costs required to deliver the comminuted biomass were significantly lower than any of the least cost fleet and equipment configurations identified with 66  the simulation model. This is largely due to the absence of any traffic congestion and the assumed 100% productivity of all equipment in the excel calculation. The resulting cost of $52.51/ODT in the I/O model was 27% less expensive than the simulated HP least cost fleet configuration of 1 operation and 6 trucks which had a total delivered cost of $72.08/ODT. This could also be interpreted as an inherent maximum efficiency of 73% given the parameters specified in the simulation model.  Table 14 Cost Breakdown of Operations by Simulation and Input/Output Model   Least Cost Simulation of HP Fixed Fleet Input / Output Model Summary  Total Trucking Time (hours) 7575 6564 Total Grinding & Loading Time (hours) 1013 1004   Total Trucking Costs ($/ODT) $40.83 $32.57 Total Grinding & Loading Costs ($/ODT) $28.60 $19.05 Total Relocation Costs ($/ODT) $2.65 $0.88 Total Cost ($/ODT) $72.08 $52.51    67  6 Discussion A number of trends and system behaviors were identified by running hundreds of iterative simulations of the biomass operations analyzed here. Some of these results were anticipated while others, while being intuitively logical, were not necessarily expected.  6.1 Sensitivity in Aggregate In calibrating the initial system for on-site diesel grinding it became apparent that the most efficient number of trucks sent to service a cutblock was dependent on the cycle time, and to an extent, the travel distance. The Dispatch Delay between trucks departing for their destination cutblock is an important factor in improving operational efficiency. This Dispatch Delay should be approximately the same as the amount of time as a truck takes between arriving and departing from a cutblock. For this system it was found that there was a fairly strong relationship between the least cost fleet size and the average cycle time, where longer cycle times require additional trucks to maintain efficient grinder and loader utilization. When identifying a least cost fleet size servicing a grinding and loading operation the iterative approach of adding one truck at a time revealed interesting trends in the broken down costs. Starting with 3 trucks servicing a distant operation, the costs per tonne are very high for the grinder and loader, but very low for transportation. This is due to near 100% efficiency for the trucks but greatly underutilized on-site equipment. As the number of trucks increases the trucking costs per ODT begin to rise while the grinding and loading costs fall. The shape of these curves is dependent on the cost structures of the equipment in use. If equipment costs are billed at flat hourly rates regardless of whether or not they are being utilized, the cost curves will be different that those generated in these simulations where discounted costs were applied while equipment would likely be shut down and therefore not consuming fuel. This difference in 68  potential cost structures represents the difference between a contractor based model of forest operations versus owning and operating equipment with the perspective of fixed and variable costs. The results of the fleet planning analysis showed that the least cost configuration occurred when the grinder and loader maximized their productive time but without incurring a backlog of trucks idling, waiting for their turn to be loaded. It was found that the relocation costs were relatively unaffected and insignificant once the equipment type and cutblock locations and volumes were set. However when block groups are analyzed in isolation, groups with less available volume tend to have greater relocation costs per ODT.  By varying the assumed biomass volume across the system it became apparent that this variable can have a significant impact on delivered cost. With even a 10% variation in the assumed biomass volume the cost impact was greater than adding or removing a truck from the least cost configuration. By increasing the available volume the delivered cost decreased (Table 12).  The block group analysis reveals great differences in delivered costs from one group to the next. These costs are largely attributed to varying transportation cost. A trend of increasing costs relative to increasing cycle time was clearly identified (Figure 10). By analyzing total delivered cost curves for the block groups it become apparent that average cycle time and biomass volume are both contributing factors to identifying least cost fleet size. It was also apparent that identifying a least cost fleet size is a more significant factor in some groups than others. It is expected that costs of processing block groups with lower volumes at greater travel distances are less dependent on identifying a least cost fleet size. These block groups tend to have flatter cost to fleet size graphs (Figure 9). This effect is likely muted due to 69  the rapid completion of operations. Before a backlog of trucks can occur, all the biomass is delivered and the simulation completes. Scheduling of variable fleet configurations could pose complications in real life operations. If a single operation is being conducted then some drivers would need to be employed periodically when operations move between smaller and larger fleet operations. This may not be desirable. By running simultaneous operations drivers could be moved between operations and attempts could be made to maintain a consistent work force. When comparing different types of equipment for processing and transporting forest harvest residues it became apparent that the higher productivity machinery was more cost effective. Although the hourly cost is significantly lower with the less productive equipment, the discount is not enough to overcome the drop in productivity. The smaller equipment had roughly a 25% cost discount but was only a little more than half as productive. This lead to operations taking approximately twice as many operating days to complete, resulting in significantly more billing hours and a 36% cost increase in the least cost fleet configuration. Aside from the cost and productivity differences, changing equipment types showed little other difference in operations within this system. One potential impact of varying equipment types that was not investigated is cutblock accessibility, with the larger equipment potentially unable to access all the biomass available to the smaller equipment. But this would likely have a greater impact on truck access than grinding and loading equipment as most cutblocks are reasonably accessible by standard lowbeds. By replacing in-block diesel grinders with centralized electric grinders it was found that costs can be greatly reduced under certain circumstances. By eliminating the expensive diesel grinders at the cutblock the operating costs were greatly reduced. Less significantly, there was 70  also a reduction in relocation costs. However, transporting unprocessed harvest residues with a much lower material density required a significantly greater number of truck loads. In addition, a second loader was required at the reload site to place the residues into the electric grinder. And while the diesel grinder costs were eliminated, costs for the electric grinder were incurred, at much lower rates however. The result was a shifting of the balance between transportation and operations costs.  These results are also heavily dependent on the assumptions around equipment type and cost allocation. While electric grinder has a lower hourly cost, the assumption that only part of its costs are attributed to the biomass grinding operations is specific to this case study and may not always be the case. Furthermore, it was assumed that harvest residues could be transported by trucks of the same capacity as hog fuel transporters. This assumption may require specialized vehicles to be valid, as existing research trials on residue transport often employ existing hook-lift truck configurations with lower capacities. Biomass densities are also likely to vary in reality, particularly when transporting unprocessed residues. The density assumptions of residues being 145 kg/m3 on average may be appropriate, but in the case of large obtusely shaped material, loading trucks may pose a significant operational challenge. In some cases it may be necessary to pre-process the harvest residues if centralized grinding were anticipated, such as the residue bundling practices commonly found in Europe. In general it was found that block groups with shorter cycle times would be more cost effective employing centralized grinding, while block groups requiring longer hauls were more economically processed in-situ. There was however no definitive threshold where blocks should use diesel versus electric grinders. Other factors, such as the available volume of material at a 71  cutblock, are important in determining the most cost effective manner of procuring biomass. While in reality block groups would need to be assessed on a case by case basis, within this system it was found that beyond approximately a four and a half hour cycle time the use of diesel grinders may be more cost effective. The shifting balance between transportation and operating costs with centralized grinders also altered the identified least cost fleet configuration. Whereas with diesel operations adding an additional operation to the system resulted in a slight increase in costs due to increased traffic congestion, the same trend was not seen with electric grinding. With the reduced cost of operations at the cutblocks more importance is placed on the transportation costs. And with loads of lower density material, trucks were modeled to be able to load and unload more quickly reducing bottlenecks at Silverdale. This resulted in a least cost fleet configuration consisting of 3 operations being serviced by a total of 17 trucks in the fixed fleet analysis.  The moisture content of harvest residues impacts the resulting fuel quality of the biomass fuel, and was also found to have an impact on operations. Initially the material moisture content was studied to help identify a strategy to deliver biomass at 30% moisture content as specified by the power generator. This was achieved in 4 of the moisture content sensitivity cases: Constant 20, Constant 30, Reduced, and 2nd Year Post Harvest. While the constant moisture content cases are not very useful from a strategic perspective, the variations in the initial moisture content schedule indicate that this biomass quality parameter can be improved through planning. As identified in the introductory discussion, it is likely that once harvest residues are given the opportunity to season for 1 year after initial timber harvest their moisture content will fall lower during the second summer season. Given this understanding, if operational and time constraints permit detailed scheduling, a moisture content of 30% can be achieved by conducting operations 72  during the summer and early fall of the second year post timber harvest. Even if operations must continue outside of this range, an average delivered moisture content below 30% can be achieved if the bulk of operations occur during this period. The assumption that the initial moisture content schedule is suitable for this analysis is based on minimal regional data and a best fit model developed for climactically similar, but distant part of the world. While there may be significant variation from one site to another, and the average moisture content may be higher as represented by the elevated input schedule, it is still expected that this general trend will persist. In reality, for this specific case study region it may only be possible to achieve an average delivered moisture content of around 42% as found with the elevated case. However, if biomass operations target the summer and fall of the second year post harvest, they will likely achieve a lower delivered moisture content. In addition to reducing moisture content to meet the demands of the power producer it was also identified that there can be certain operational benefits to transporting dryer material. Transport trucks are limited by the weight they can carry as well as their volume. When the moisture content of the material their transporting reaches a certain threshold they begin to hit their weight limit before filling up volumetrically. This results in an increased number of truck loads required to transport the same amount of biomass, which leads to an increase in operating and transportation costs. For the trucks and assumed hog fuel density used in this case study the theoretical threshold occurs at 31.5% moisture content. This cost impact can be quite significant. Where material delivered below the threshold was found to have an average cost of $67.95/ODT, using the elevated moisture content input schedule resulted in an increase of $8.74/ODT and the Constant 50% schedule had an increase of $19.16/ODT. These potential variations in operational efficiency through managing moisture therefore become a significant cost consideration. 73  6.2 Varying Complexity of Methods There are a range of methods available to those planning and analyzing supply chains. The initial method for this case study used a complex simulation model allowing for detailed analysis of product quality parameters, equipment interactions, traffic, operations, and scheduling. The second method employed was a simplified spreadsheet input/output model resulting in fast production of results but allowing for little understanding of the systems internal processes.  The simulation modeling allowed for the real time visualization of activities within the case study system. As a result, many observations were made that would only otherwise have been possible by monitoring real life operations. These included adjustments to truck scheduling to improve equipment utilization and the addition of pullouts on resource roads to improve traffic flows. Tracking material moisture content was a simple addition to the simulation model and was critical in identifying the unanticipated operational and cost impacts it has on the system. Once developed, the simulation model was easily manipulated to explore sensitivities in input parameters as well as querying alternative operational configurations. By analyzing block groups individually with both conventional diesel grinding and centralized electric grinding a relationship was established identifying what the most cost effective form of processing would be on a sub-regional basis. While average cycle time may have been thought to be the only determining factor, the simulation results indicated that parameters such as available block volume also have an impact. Though the complexity of simulation modeling has significant advantages in understanding system dynamics, the basic spreadsheet analysis is not without value. The simulation model required significant calibration to emulate real world operations. The same 74  would also be true with the spreadsheet, but the process could be as simple as scaling all operations to an observed efficiency. This approach would not allow for the identification of unintended system interactions, and would not hold constant if applied to an alternate system. The questions of scale and end use become significant when identifying an appropriate method. While the simulation model becomes ever more complex and more difficult to calibrate with scale, the use of a spreadsheet with aggregated  data inputs and results will likely become more accurate the larger the system under scrutiny. However this will lead to a less informative analysis and not provide the insight into system dynamics and interaction of components visible through simulation modeling.   75  7 Conclusion In order to gain an understanding of the operational factors impacting forest harvest residue collection operations, a case study set in coastal British Columbia was examined. A simulation model of the case study area was developed to conduct the analysis. Expected variations in biomass moisture content for the region were also determined, and incorporated in the model. 7.1 Reducing Costs Through developing, calibrating and performing sensitivity analysis with the AROMA simulation model, a thorough understanding of the case study system was achieved. These processes and their results helped identify a number of potential cost saving considerations for forest biomass operations that would be possible to implement prior to the start of operations. These included the ability to identify the need and approximate location for pullouts on resource roads, helped identify an efficient fleet size, and specify appropriate equipment configurations.  The sensitivity analysis helped illustrate the importance of accurate data to the operational planning process. Data collection methods employed in this analysis were based largely around publicly available information. With greater site specific detail that would be available to actual operators, the inputs and assumptions could likely be improved upon. And the continuous collection of operational data would help further refine the analysis for better future planning. Regardless, several parameters were identified which result in significant cost implications. Fleet size is an important factor in reducing the delivered cost of biomass, but so long as the number of trucks employed falls within a certain range, other factors will cause greater impacts. The range in fleet size is dependent on both average cycle time to an operation and the 76  available volume at the site. Therefore good geographic information on the site and access roads is important. Furthermore, information on the sites timber harvest and its related biomass ratio are key to assessing the available biomass. Similarly to fleet size planning, a variation in residue volume of 20% can result in nearly a $5/ODT fluctuation in the price of delivered biomass.  A potentially more significant cost impact is found when considering what equipment to employ and in what configuration it should be used. In this case study switching to low productivity machinery from high productivity machinery increased costs by greater than $25/ODT. This variation is explained not in the productivities themselves but in the ratio of hourly costs to production rates. If an operation were able to commence with half the productivity and half the cost then the choice in machinery may be less relevant. However, due to the nature of forest operations, with significant fixed costs incurred to operate it was found that the productivity of machinery is of great importance.  The option to use a centralized grinding configuration also impacts the cost of operations. Whether or not to employ centralized grinding will be largely dependent on the average cycle time to a specific operation. For this case study it was found that operations with a cycle time of less than 4.5 hours would be more cost effective when transporting unprocessed harvest residues to a centralized electric grinder. The resulting difference between potential operating costs of in-situ versus centralized grinding could be greater than $13/ODT. It is expected however that the amount of available biomass on site will also impact this cost difference so specific operations should be assessed on a case by case basis. 7.2 Moisture Content Moisture content is a significant quality factor of hog fuel as well as an important operational consideration. It was identified through a review of the relevant literature that 77  logging residues in the coastal British Columbia environment are likely lower after being piled on site for two summers rather than one. By incorporating this information into planning of biomass operations a higher quality of hog fuel can be procured.  It was also identified through simulation modeling that the moisture content of harvest residues becomes an important cost consideration. When the moisture content of biomass is over a certain threshold trucks become limited by weight rather than volume capacities. For the High Productivity trucks used in this case study that threshold was found to be 31.5%. Beyond this point more trips are required to transport the same amount of material thereby increasing the costs. By conducting operations in the first summer post timber harvest rather than the second, a cost increase of greater than $4/ODT was identified. And transporting biomass with 50% moisture content compared with the Second Year Post Harvest scenario could increase costs by nearly $20/ODT. In light of the benefits of processing harvest residues in the second year after timber harvest, it’s important that the obligations of timber harvesters are not contrary to this objective. Current regulations dictate that harvest residue piles on crown land in BC should be burnt or processed in the first year after timber harvest to allow for tree regeneration, and reduce wild fire hazards. In light of these findings, amending this policy to allow biomass processors more flexibility in scheduling operations may be justified. 7.3 Benefits of Simulation Simulation modeling has several advantages over simplified spreadsheet approaches, but only if the problem being analyzed requires more detailed systems analysis. Understanding issues around equipment and vehicle coordination is made possible by analyzing interactions between the autonomous entities in the simulation environment. Tracking quality factors such as 78  moisture content would be possible within an input/output spreadsheet model approach, but would require additional programing and would not present unexpected results such as the issue of partially loaded trucks. Simulation modeling is therefore a crucial tool to be implemented in understanding and fine tuning system dynamics. But if all that’s required in an analysis is aggregated total values for parameters such as time or cost, a simple spreadsheet approach may be more suitable. As with any analysis, the methods employed should be determined by the questions being posed. 7.4 Future Work Numerous assumptions had to be made to conduct this analysis. In order to further refine the results and produce a more valuable analysis tool additional data collection would be a priority. Better information on harvest residue moisture content for BC’s coast would help refine a strategy to manage biomass quality. A more detailed assessment of vehicle types and access would enable more specific determinations of which cutblocks could be feasibly processed, and what operational configurations should be employed. Finally, assessing alternative markets for harvest residues, such as torrefied pellets or petrochemical substitutes may increase the demand for these resources and increase the economic feasibility of their collection.  79  Bibliography Acuna, M., Anttila, P., Sikanen, L., Prinz, R., & Asikainen,  a. (2012). Predicting and controlling moisture content to optimise forest biomass logistics. Croatian Journal of Forest Engineering, 33, 225–238. Akanle, O. M., & Zhang, D. Z. (2008). Agent-based model for optimising supply-chain configurations. International Journal of Production Economics, 115(2), 444–460.  Alakangas, E. (2005). Properties of wood fuels used in Finland – BIOSOUTH -project. Bulk Solids Handling. Alam, M. B., Pulkki, R., & Shahi, C. (2012). Woody biomass availability for bioenergy production using forest depletion spatial data in northwestern Ontario. Canadian Journal of Forest Research, 42(3), 506–516.  Alam, M. B., Pulkki, R., Shahi, C., & Upadhyay, T. (2012). Modeling woody biomass procurement for bioenergy production at the Atikokan Generating station in northwestern Ontario, Canada. Energies, 5(12), 5064–5085.  Alam, M., Pulkki, R., & Shahi, C. (2012). Road network optimization model for supplying woody biomass feedstock for energy production in northwestern Ontario. The Open Forest Science Journal, 5, 1–14.  Alfieri, A., & Brandimarte, P. (1997). Object-oriented modeling and simulation of integrated production/distribution systems. Computer Integrated Manufacturing Systems, 10(4), 261–266.  Amaran, S., Sahinidis, N. V, Sharda, B., & Bury, S. J. (2014). Simulation optimization: a review of algorithms and applications. 4or, 12(4), 301–333.  Amini, M., Wakolbinger, T., Racer, M., & Nejad, M. G. (2012). Alternative supply chain 80  production-sales policies for new product diffusion: An agent-based modeling and simulation approach. European Journal of Operational Research, 216(2), 301–311.  An, H., Wilhelm, W. E., & Searcy, S. W. (2011). Biofuel and petroleum-based fuel supply chain research: A literature review. Biomass and Bioenergy, 35(9), 3763–3774.  Asikainen, A. (2001). Simulation of logging and barge transport of wood from forests on islands. International Journal of Forest Engineering, 12(2), 43–50.  Asikainen, A. (2010). Simulation of stump crushing and truck transport of chips. Scandinavian Journal of Forest Research, 25(3), 245–250.  Audy, J., & Rönnqvist, M. (2012). Planning methods and decision dupport systems in vehicle routing problems for timber transportation : A review. Cirrelt, 45. Baxter, G. (2008). Assessing moisture content of piled woody debris : implications for burning. FPInnovations-Feric Report, 1–15. BC Government. (2013). DataBC. Retrieved July 22, 2013, from https://data.gov.bc.ca/ BC Ministry of Forests. (2013). Harvest Billing System. Retrieved July 22, 2013, from https://a100.gov.bc.ca/pub/hbs/ BC Office of the Premier. (2008). News release: New Bioenergy Strategy Advances Innovation, (416), 33–35. Beaudoin, D., Frayret, J.-M., & LeBel, L. (2008). Hierarchical forest management with anticipation: an application to tactical–operational planning integration. Canadian Journal of Forest Research, 38(8), 2198–2211.  Becker, M., Wenning, B.-L., Görg, C., Gehrke, J. D., Lorenz, M., & Herzog, O. (2006). Agent-based and discrete event simulation of autonomous logistic processes. Proceedings of: Borutzky, W.; Orsoni, A.; Zobel, R.(eds.), 571, 566–571. 81  Biswas, S., & Narahari, Y. (2003). Object oriented modeling and decision support for supply chains. European Journal of Operational Research, 153(3 SPEC. ISS.), 704–726.  Bosona, T. G., & Gebresenbet, G. (2011). Cluster building and logistics network integration of local food supply chain. Biosystems Engineering, 108(4), 293–302.  Bousquet, F., Barreteau, O., Le Page, C., Mullon, C., & Weber, J. (1999). An environmental modelling approach. The use of multi-agent simulations. Advances in Environmental and Ecological Modelling. Briggs, D. (1994). Forest products measurements and conversion factors: With special emphasis on the US Pacific Northwest Chapter 7. Chapter 7. Chips, Sawdust, Planer Shavings, Bark, and Hog Fuel. University of Washington Institute of Forest Resources, AR-10, Seattle, Washington 98195.  Brown, G. (2006). Optimizing the US Navy’s Combat Logistics Force, Wiley Periodicals Naval Research and Logistics, 55: 800-810 Cambero, C., Sowlati, T., & Pavel, M. (2015). Economic and life cycle environmental optimization of forest-based biorefinery supply chains for bioenergy and biofuel production. Chemical Engineering Research and Design, 1–18.  Caputo, A. C., Pelagagge, P. M., & Scacchia, F. (2003). Integrating transport systems in supply chain management software tools. Industrial Management & Data Systems, 103(7), 503–515.  Carlsson, D., & Rönnqvist, M. (2005). Supply chain management in forestry––case studies at Södra Cell AB. European Journal of Operational Research, 163, 589–616.  Caunhye, A. M., Nie, X., & Pokharel, S. (2012). Optimization models in emergency logistics: A literature review. Socio-Economic Planning Sciences, 46(1), 4–13.  82  Chatfield, D. C., Harrison, T. P., & Hayya, J. C. (2003). SISCO: The simulator for integrated supply chain operations, 1–41. Chen, E. J., Lee, Y. M., & Selikson, P. L. (2002). A simulation study of logistics activities in a chemical plant. Simulation Modelling Practice and Theory, 10(3-4), 235–245.  Chopra, S., & Meindl, P. (2010). Supply Chain Management: Strategy, Planning, and Operation. Pearson Education. Cross, J. C., Turnblom, E. C., & Ettl, G. J. (2013). Biomass production on the Olympic and Kitsap Peninsulas, Washington: Updated logging residue ratios, slash pile volume-to-weight ratios, and supply curves for selected locations. USDA Forest Service - General Technical Report PNW-GTR, (872), 1–30.  Dal-Mas, M., Giarola, S., Zamboni, A., & Bezzo, F. (2011). Strategic design and investment capacity planning of the ethanol supply chain under price uncertainty. Biomass and Bioenergy, 35(5), 2059–2071.  Davidsson, P., Johansson, S. J., Persson, J. a, & Wernstedt, F. (2003). Agent-based Approaches and Classical Optimization Techniques for Dynamic Distributed Resource Allocation : A preliminary study. AAMAS’03 Workshop on Representations and Approaches for TimeCritical Decentralized Resource/Role/Task Allocation. De Mol, R. M. d., Jogems, M. a. H., Beek, P. Van, & Gigler, J. K. (1997). Simulation and optimization of the logistics of biomass fuel collection. NJAS Wageningen Journal of Life Sciences.  Dyson, P. (2013). Effects of Season on the Moisture Content of Roadside Harvesting Debris in Coastal British Columbia, FPInnovations Technical Report, 1-13. Ebadian, M., Sowlati, T., Sokhansanj, S., Stumborg, M., & Townley-Smith, L. (2011). A new 83  simulation model for multi-agricultural biomass logistics system in bioenergy production. Biosystems Engineering, 110(3), 280–290.  Ebadian, M., Sowlati, T., Sokhansanj, S., Townley-Smith, L., & Stumborg, M. (2013). Modeling and analysing storage systems in agricultural biomass supply chain for cellulosic ethanol production. Applied Energy, 102, 840–849.  Eisentraut, A., & Brown, A. (2012). Technology Roadmap Bioenergy for Heat and Power. Technology Roadmaps, (2), 1–41.  Esteban, L. S., & Carrasco, J. E. (2011). Biomass resources and costs: Assessment in different EU countries. Biomass and Bioenergy, 35, S21–S30.  Evans, A. M. (2008). Synthesis of knowledge from woody biomass removal case studies, 39. Farve, R., & Napper, C. (2009). Biomass fuels & whole tree harvesting impacts on soil productivity—Review of literature, US Forest Service National Technology and Development Program Report - Inventory and Modeling 0920 1803-SDTDC, 66.  Fleming, R. L., Powers, R. F., Foster, N. W., Kranabetter, J. M., Scott, D. A., Ponder Jr., F., … Tiarks, A. E. (2006). Effects of organic matter removal, soil compaction, and vegetation control on 5-year seedling performance: a regional comparison of Long-Term Soil Productivity sites. Canadian Journal of Forest Research, 36(3), 529–550.  Forrester, J. W. (1961). Industrial dynamics. Harvard Business Review. Fraser Valley Guide. (2015). HISTORY -- VALLEY REGION -- FRASER VALLEY GUIDE- Settlement by First Nations and recent development in the Fraser Valley, BC, Canada. Retrieved on January 15, 2016 from http://www.fraservalleyguide.com/History.html Frayret, J. M., D’Amours, S., Rousseau,  a., Harvey, S., & Gaudreault, J. (2007). Agent-based supply-chain planning in the forest products industry. International Journal of Flexible 84  Manufacturing Systems, 19(4), 358–391.  Gan, J. (2007). Supply of biomass, bioenergy, and carbon mitigation: Method and application. Energy Policy, 35(12), 6003–6009.  Gautam, S. H., Beaudoin, D., Gautam, S. H., Lebel, L., Beaudoin, D., & Simard, M. (2014). Examining the Long-Term Impact of Operational Level Silvicultural Flexibility in a Hierarchical Planning Framework : A Simulation-Optimization Based Approach Examining the Long-Term Impact of Operational Level Silvicultural Flexibility in a Hierarchical Planning Framework. Cirrelt-2014-62 (November). Gautam, S., Pulkki, R., Shahi, C., & Leitch, M. (2012). Fuel quality changes in full tree logging residue during storage in roadside slash piles in Northwestern Ontario. Biomass and Bioenergy, 42, 43–50.  Gautier, A., Lamond, B. F., Paré, D., & Rouleau, F. (2000). The Québec ministry of natural resources uses linear programming to understand the wood-fiber market. Interfaces, 30(6), 32–48.  Gehrke, J. D., & Ober-Blöbaum, C. (2007). Multiagent-based Logistics Simulation with PlaSMA. Simulation, 1(107), 1–4.  Ghodsypour, S. H., & O’Brien, C. (2001). The total cost of logistics in supplier selection, under conditions of multiple sourcing, multiple criteria and capacity constraint. International Journal of Production Economics, 73(1), 15–27. Govett, R., Mace, T., & Bowe, S. (2010). A Practical Guide for the Determination of Moisture Content, Wisconcin Department of Natural Resources. 1–20. Gronalt, M., & Rauch, P. (2007). Designing a regional forest fuel supply network. Biomass and Bioenergy, 31(6), 393–402.  85  Hakkila, P. (1989). Utilization of Residual Forest Biomass. Springer Series in Wood Science. 581 Hall, P., Gigler, J. K., & Sims, R. E. H. (2001). Delivery systems of forest arisings for energy production in New Zealand. Biomass and Bioenergy, 21(6), 391–399.  Jirjis, R. (1995). Storage and drying of wood fuel. Biomass and Bioenergy, 9(1-5), 181–190.  Johansson, J., Liss, J. E., Gullberg, T., & Björheden, R. (2006). Transport and handling of forest energy bundles - Advantages and problems. Biomass and Bioenergy, 30(4), 334–341.  Jones, G., Loeffler, D., Butler, E., & Hummel, S. (2009). Emissions , Energy Return and Economics From Utilizing Forest Residues for Thermal Energy Compared to Onsite Pile Burning. Integrated Management of Carbon Sequestration and Biomass Utilization Opportunities in a Changin Climage: Proceedings of the 2009 National Silviculture Workship, 145–158. Joseph Buongiorno. (2003). Decision Methods for Forest Resource Management. Zhurnal Eksperimental’noi i Teoreticheskoi Fiziki.  Junginger, M., Faaij,  a., Björheden, R., & Turkenburg, W. C. (2005). Technological learning and cost reductions in wood fuel supply chains in Sweden. Biomass and Bioenergy, 29(6), 399–418.  Kärhä, K. (2011). Industrial supply chains and production machinery of forest chips in Finland. Biomass and Bioenergy, 35(8), 3404–3413.  Kärhä, K., & Vartiamäki, T. (2006). Productivity and costs of slash bundling in Nordic conditions. Biomass and Bioenergy, 30(12), 1043–1052.  Kerstetter, J. D., & Lyons, J. K. (2001). Logging and Agricultural Residue Supply Curves for the Pacific Northwest Acknowledgments. US Department of Energy, (January). 86  King, D. H., & Harrison, H. S. (2013a). Open Source Simulation Software ―JAAMSIM.‖ In Proceedings - Winter Simulation Conference (pp. 2163–2171).  King, D. H., & Harrison, H. S. (2013b). Open Source Simulation Software ―JAAMSIM.‖ Proceedings - Winter Simulation Conference, 2163–2171.  Kinoshita, T., Inoue, K., Iwao, K., Kagemoto, H., & Yamagata, Y. (2009). A spatial evaluation of forest biomass usage using GIS. Applied Energy, 86(1), 1–8.  Kleijnen, J. P. C. (1997). Sensitivity analysis and related analyses: A review of some statistical techniques. Journal of Statistical Computation and Simulation (Vol. 57).  Klose, A., & Drexl, A. (2005). Facility location models for distribution system design. European Journal of Operational Research, 162, 4–29. http://doi.org/10.1016/j.ejor.2003.10.031 Klügl, F., Herrler, R., & Fehler, M. (2006). SeSAm : Implementation of Agent-Based Simulation Using Visual Programming, AAMAS 06, May 18-12, ACM. 1–2. Kocoloski, M., Michael Griffin, W., & Scott Matthews, H. (2011). Estimating national costs, benefits, and potential for cellulosic ethanol production from forest thinnings. Biomass and Bioenergy, 35(5), 2133–2142.  Kumar, A., Flynn, P. C., & Sokhansanj, S. (2005). Feedstock Availability and Power Costs Associated with Using BC’s Beetle-Infested, BIOCAP Canada. 1–49. Law, A. M. (2009). How to build valid and credible simulation models. Proceedings - Winter Simulation Conference, 24–33.  Lehtikangas, P. (2001). Quality properties of pelletised sawdust, logging residues and bark. Biomass and Bioenergy, 20(5), 351–360. Lehtonen, J.-M., & Seppala, U. (1997). A methodology for data gathering and analysis in a logistics simulation project. Integrated Manufacturing Systems, 8(6), 351–358.  87  Lindroos, O., & Sowlati, T. (2011). Costs, CO 2 Emissions, and Energy Balances of Applying Nordic Slash Recovery Methods in British Columbia, West. J. Appl. For. 26(1). Lyons, K. (2013). personal communication - June 20, 2013. MacDonald, A. J. (2009). Assessment of economically accessible biomass. BC Coastal Forest Sector Hem-Fir Initiative Technical Report, (May), 18 pp. Macdonald, A. J. (2012). Assessment of Forest Feedstock ( Biomass ) for Campbell River by February 2012 Prepared for : BC Ministry of Jobs , Tourism and Skills Training, (February). Manuj, I., Mentzer, J. T., & Bowers, M. R. (2009). Improving the rigor of discrete-event simulation in logistics and supply chain research. International Journal of Physical Distribution & Logistics Management, 39(3), 172–201.  Maoh, H., Kanaroglou, P., & Woudsma, C. (2008). Simulation Model for Assessing the Impact of Climate Change on Transportation and the Economy in Canada. Transportation Research Record: Journal of the Transportation Research Board, 2067(-1), 84–92.  McIver, I. (2013a). personal communication - July 12, 2013. McIver, I. (2013b). personal communication - July 23, 2013. McIver, I. (2013c). personal communication - September 12, 2013. McKendry, P. (2002). Energy production from biomass (part 1): Overview of biomass. Bioresource Technology, 83(1), 37–46.  Miles, P. D., & Smith, W. B. (2009). Specific Gravity and Other Properties of Wood and Bark for 156 Tree Species Found in North America. Res. Note. NRS-38, 35. Mobini, M. (2015). On the design and analysis of forest biomass to biofuel and bioenergy supply chains, (April). 88  Mobini, M., Meyer, J. C., Trippe, F., Sowlati, T., Fröhling, M., & Schultmann, F. (2014). Assessing the integration of torrefaction into wood pellet production. Journal of Cleaner Production, 78, 216–225.  Mobini, M., Sowlati, T., & Sokhansanj, S. (2011). Forest biomass supply logistics for a power plant using the discrete-event simulation approach. Applied Energy, 88(4), 1241–1250.  Mobini, M., Sowlati, T., & Sokhansanj, S. (2013). A simulation model for the design and analysis of wood pellet supply chains. Applied Energy, 111, 1239–1249.  Morey, R. V, Kaliyan, N., Tiffany, D. G., & Schmidt, D. R. (2010). A Corn Stover Supply Logistics System, 26(3), 455–462. Morneau-Pereira, M., Arabi, M., Gaudreault, J., Nourelfath, M., & Ouhimmou, M. (2013). An Optimization and Simulation Framework for Integrated Tactical Planning of Wood Harvesting Operations, Wood Allocation and Lumber Production, Cirrelt. 16.  Murphy, F., Devlin, G., & McDonnell, K. (2014). Forest biomass supply chains in Ireland: A life cycle assessment of GHG emissions and primary energy balances. Applied Energy, 116, 1–8.  Naik, S., Goud, V., Rout, P., & Dalai, A. (2010). Production of first and second generation biofuels: a comprehensive review. Renewable and Sustainable Energy Reviews, 14(2), 578–597.  Niquidet, K., Stennes, B., & van Kooten, G. C. (2012). Bioenergy from Mountain Pine Beetle Timber and Forest Residuals: A Cost Analysis. Canadian Journal of Agricultural Economics, 60(2), 195–210.  Norum, R. a., & Miller, M. (1984). Measuring Fuel Moisture Content in Alaska : Standard Methods and. General Technical Report, 1–40. 89  Nurmi, J. (1999). The storage of logging residue for fuel. Biomass and Bioenergy, 17(1), 41–47.  Palander, T. S., & Voutilainen, J. J. (2013). Modelling fuel terminals for supplying a combined heat and power (CHP) plant with forest biomass in Finland. Biosystems Engineering, 114(2), 135–145.  Panichelli, L., & Gnansounou, E. (2008). GIS-based approach for defining bioenergy facilities location: A case study in Northern Spain based on marginal delivery costs and resources competition between facilities. Biomass and Bioenergy, 32, 289–300.  Paré, D., Bernier, P., Thiffault, E., & Titus, B. D. (2011). The potential of forest biomass as an energy supply for Canada. The Forestry Chronicle, 87(1), 71–76.  Perrin, R., Sesmero, J., Wamisho, K., & Bacha, D. (2012). Biomass supply schedules for Great Plains delivery points. Biomass and Bioenergy, 37, 213–220.  Pettersson, M., & Nordfjell, T. (2007). Fuel quality changes during seasonal storage of compacted logging residues and young trees. Biomass and Bioenergy, 31(11-12), 782–792.  Praça, I., Ramos, C., Vale, Z., & Cordeiro, M. (2003). Mascem: A Multiagent System that Simulates Competitive Electricity Markets. IEEE Intelligent Systems, 18, 54–60.  Ralevic, P., Ryans, M., & Cormier, D. (2010). Assessing forest biomass for bioenergy: Operational challenges and cost considerations. Forestry Chronicle, 86(1), 43–50.  Ranta, T., & Korpinen, O. (2011). How to analyse and maximise the forest fuel supply availability to power plants in Eastern Finland. Biomass and Bioenergy, 35(5), 1841–1850.  Rauch, P., & Gronalt, M. (2011). The effects of rising energy costs and transportation mode mix on forest fuel procurement costs. Biomass and Bioenergy, 35(1), 690–699.  Rönnqvist, M. (2003). Optimization in forestry. Mathematical Programming, 97(1-2), 267–284.  Röser, D., Mola-Yudego, B., Sikanen, L., Prinz, R., Gritten, D., Emer, B., … Erkkilä, A. (2011). 90  Natural drying treatments during seasonal storage of wood for bioenergy in different European locations. Biomass and Bioenergy, 35(10), 4238–4247.  Rossetti, M. D., & Chan, H. (2003). A Prototype Object-Oriented Supply Chain Simulation Framework. Proceedings of the 2003 Winter Simulation Conference, (1991), 1612–1620.  Shabani, N., & Sowlati, T. (2013). A mixed integer non-linear programming model for tactical value chain optimization of a wood biomass power plant. Applied Energy, 104, 353–361.  Shannon, R. E. (1998). Introduction to the art and science of simulation. 1998 Winter Simulation Conference. Proceedings (Cat. No.98CH36274), 1, 7–14.  Shen, W. (2006). IT for Balanced Manufacturing Systems: Collection of papers. IFIP TC5, WG 5.5 Seventh International Conference on Information Technology for Balanced Automation Systems in Manufacturing and Services, Niagra Falls, Ontario, Canada, September 4-6, 2006, (1), 1–5.  Sikanen, L. (2010). Effect of moisture content on the financial and environmental cost of forest biomass. European Cooperative in Science and Technology 2010 November Conference Proceedings. Sikanen, L., Röser, D., Anttila, P., & Prinz, R. (2012). Forecasting Algorithm for Natural Drying of Energy Wood in Forest Storages. Forest Energy Observer, Study Report 27, 358(0). Sims, R. E. H., & Venturi, P. (2004). All-year-round harvesting of short rotation coppice eucalyptus compared with the delivered costs of biomass from more conventional short season, harvesting systems. Biomass and Bioenergy, 26(1), 27–37.  Smith, D. (2010). WOODY BIOMASS Blitz grinder. Oregon State University, Wood Science and Engineering. 8–9. Sokhansanj, S., Kumar, A., & Turhollow, A. F. (2006). Development and implementation of 91  integrated biomass supply analysis and logistics model (IBSAL). Biomass and Bioenergy, 30(10), 838–847.  Sokhansanj, S., Turhollow, A. F., Stephen, J., Stumborg, M., Fenton, J., & Mani, S. (2008). Analysis of five simulated straw harvest scenarios. Canadian Biosystems Engineering, 50(2), 27–37. Son, Y. Il, Yoon, S. J., Kim, Y. K., & Lee, J. G. (2011). Gasification and power generation characteristics of woody biomass utilizing a downdraft gasifier. Biomass and Bioenergy, 35(10), 4215–4220.  Soulié, J. C., & Thébaud, O. (2006). Modeling fleet response in regulated fisheries: An agent-based approach. Mathematical and Computer Modelling, 44, 553–564.  Spinelli, R., & Magagnotti, N. (2009). Logging residue bundling at the roadside in mountain operations. Scandinavian Journal of Forest Research, 24(2), 173–181.  Stasko, T. H., Conrado, R. J., Wankerl, A., Labatut, R., Tasseff, R., Mannion, J. T., … Knott, G. (2011). Mapping woody-biomass supply costs using forest inventory and competing industry data. Biomass and Bioenergy, 35(1), 263–271.  Stefanovic, D., Stefanovic, N., & Radenkovic, B. (2009). Supply network modelling and simulation methodology. Simulation Modelling Practice and Theory, 17(4), 743–766.  Stennes, B. K., McBeath,  a., & Centre, P. F. (2006). Bioenergy options for woody feedstock: are trees killed by mountain pine beetle in British Columbia a viable bioenergy resource? Forestry. Natural Resources Canada, CFS, Pacific Forestry Center Information Report. BC-X-405 Tahvanainen, T., & Anttila, P. (2011). Supply chain cost analysis of long-distance transportation of energy wood in Finland. Biomass and Bioenergy, 35(8), 3360–3375.  92  Tako, A. a., & Robinson, S. (2012). The application of discrete event simulation and system dynamics in the logistics and supply chain context. Decision Support Systems, 52(4), 802– Towill, D. R., Naim, M. M., & Wikner, J. (1993). Simulation Models in the Design of Supply Chains. International Journal Of Physical Distribution. Vol. 22 Iss 5 pp. 3-13 United Nations. (2015). Adoption of the Paris Agreement. Conference of the Parties on Its Twenty-First Session, 21932(December), 32.  Upadhyay, T. P., Shahi, C., Leitch, M., & Pulkki, R. (2012). Economic feasibility of biomass gasification for power generation in three selected communities of northwestern Ontario, Canada. Energy Policy, 44, 235–244.  USEPA. (2007). Biomass Combined Heat and Power Catalog of Technologies. US Environmental Protection Agency, (September). Vahid, S. (2011). An agent-based supply chain model for strategic analysis in forestry, (December), UBC cIRcle. 1–152. Verbesselt, J., Fleck, S., Coppin, P., & Viegas. (2002). Estimation of fuel moisture content towards Fire Risk Assessment. Forest Fire Research & Wildlife Fire Safety, (Cohen 1991), 1–11. Viana, H., Cohen, W. B., Lopes, D., & Aranha, J. (2010). Assessment of forest biomass for use as energy. GIS-based analysis of geographical availability and locations of wood-fired power plants in Portugal. Applied Energy, 87(8), 2551–2560.  Viney, N. (1991). A Review of Fine Fuel Moisture Modelling. International Journal of Wildland Fire, 1(4), 215.  Wan, X., Pekny, J. F., & Reklaitis, G. V. (2005). Simulation-based optimization with surrogate models - Application to supply chain management. Computers and Chemical Engineering, 93  29(6 SPEC. ISS.), 1317–1328.  Zhang, F., Johnson, D. M., & Johnson, M. a. (2012). Development of a simulation model of biomass supply chain for biofuel production. Renewable Energy, 44, 380–391.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0300235/manifest

Comment

Related Items