Integrating MTS Commerce Data with Multimodal Freight Transportation Performance Measures to Support MTS Maintenance Investment Decision Making (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

11 S E C T I O N 2 Selection of Port Areas for Analysis The first step in developing a model that takes into account the MTS and the surface transportation system as one inte- grated system is to identify high-volume freight MTS corri- dors. By focusing on high-volume corridors, attention can be focused on projects that will have the largest potential benefit on the overall freight transportation system. The NCFRP Project 42 researchers examined publicly avail- able databases and other sources in order to identify seven pri- mary MTS corridors in the United States on the Atlantic, Gulf, and West coasts; on the Great Lakes; and on the inland water- way system. The primary metric was the total tonnage carried since the focus of this study is to facilitate freight through- put; also, tonnage information is readily available and is more accurate than available cargo value estimates. A secondary concern was to achieve geographical and cargo diversity. Using statistics from the Corps’ Navigation Data Center, the researchers ranked ports across the country in three ways: (1) by dry cargo tonnage, (2) by petroleum and petroleum products tonnage, and (3) by the number of containers moved (TEUs). These rankings were then added together to form a composite score, and the port areas with the highest rankings were selected as candidates for further study. The initial list consisted of the following: • Los Angeles/Long Beach, California • South Louisiana, Louisiana • Hampton Roads (Port of Virginia), Virginia • Philadelphia, Pennsylvania • Huntington, West Virginia • Sabine-Neches Waterway (Beaumont, Port Arthur, and Orange), Texas • Charleston, South Carolina This initial list did not include a Great Lakes port while it included three ports on the East Coast. Great Lakes ports present a different set of modeling challenges compared to other ports, so it was important to modify the list to include at least one Great Lakes port. The researchers identified the Great Lakes port with the highest ranking—Duluth, Minnesota—and substituted it for Philadelphia, Pennsylva- nia, in order to achieve better geographical diversity. Addi- tionally, the magnitude and complexity of the cargo flows through Los Angeles/Long Beach would have consumed a significant amount of project resources; therefore, the research team, in consultation with the NCFRP Project 42 panel, substituted Portland, Oregon, for Los Angeles/Long Beach, California. The availability of more current rail data and the connection to the Snake/Columbia River system were deciding factors in this selection. Finally, the Corps indicated that the data available in CPT for South Louisi- ana were inconsistent with other databases that it maintains for the same area. The Port of Plaquemines, Louisiana, has cargo and transportation characteristics similar to those of South Louisiana, although on a smaller scale, and there are fewer issues with potential inconsistencies. Therefore, the researchers substituted the Port of Plaquemines, Louisiana, for South Louisiana. The researchers presented the list to the NCFRP Project 42 panel and discussed each port. The panel selected five ports to focus on as case studies: • Duluth, Minnesota • Hampton Roads (Port of Virginia), Virginia • Huntington, West Virginia • Plaquemines, Louisiana • Portland, Oregon The Port of Portland was further subdivided into deep draft (import/export) traffic and inland water (Columbia/Snake River) traffic. Therefore, even though there are five ports on the list, there are actually six case studies. Selection and Characterization of High-Volume Freight MTS Corridors in the United States

12 Each port area was then described in terms of certain pri- mary physical characteristics. The primary criteria used to characterize each port include the following: • Type of port. Each port was classified as an inland, coastal, or Great Lakes port. Each of these classifications embod- ies a unique set of modeling characteristics. For example, Great Lakes ports work within a substantially closed system, but this system includes two countries. Inland ports can handle traffic that originates at any point on the inland waterway system and connect with a wide variety of land- based transportation networks. Coastal ports are essentially end nodes in a land transportation system for purposes of this research. • Primary commodities. The available Corps data reported commodity flows using two-digit commodity codes. The research team ranked commodity flows in descending order of average annual tonnages over the period 2006 to 2010 (the data from CPT that were provided to the project team) and selected the top-ranking commodities that together made up at least 80 percent of the total flows for a port. Duluth, Minnesota (Great Lakes) Duluth is different from the other ports included in the study because a very high percentage of its total commod- ity flows is outbound. In fact, for the primary commodities, the flows are entirely outbound. Additionally, Duluth tends to originate commodities destined for users located in other port cities on the Great Lakes. Therefore, its interaction with the surface transportation system primarily consists of inbound movements via rail. Table 1 shows the primary commodities at Duluth. Although Code 44 includes “Steel Waste & Scrap” in its title, in the case of Duluth, the tonnage in this category is composed entirely of iron ore (taconite). In Duluth, a high percentage of the waterborne shipments terminate at domes- tic locations. Therefore, when defining the commodity flows for Duluth, it is important to look at the surface transporta- tion flows into the port and the resulting outbound flows to other Great Lakes ports in the United States. The commodity flows for Duluth are analogous in many respects to the com- modity flows described for Huntington, West Virginia. Hampton Roads, Virginia (Coastal) Coal and containerized cargoes dominate the commodity mix for Hampton Roads. The detailed data provided by the Corps reported 90 percent of the actual total traffic. Table 2 shows the totals provided by the Corps and extrapolates them to the 100 percent level (internal and local traffic not included). The five commodity groups listed in Table 2 make up 81 per- cent of the total cargo flow for Hampton Roads. An examination of the data revealed that there is virtu- ally no coastwise activity for any of these commodity groups; therefore, Hampton Roads was considered to be the origin for imports and the destination for exports. Huntington, West Virginia (Inland) Huntington, West Virginia, is potentially the most com- plex of the candidate ports. This is because freight flows can originate and terminate on any number of waterway seg- ments outside of the port area proper, with potential con- nections to highway and rail links that extend even farther. Furthermore, because the research team did not have access Table 1. Primary commodity categories for Duluth, Minnesota. Commodity Category (Code) Average Annual Tonnage 2006–2010 (in 000s) Coal, Lignite & Coal Coke (10) 19,825 Iron Ore and Iron & Steel Waste & Scrap (44) 15,630 Total 35,455 Total All Commodities 40,721 All tonnage figures were provided by the Corps from its CPT unless otherwise noted. Table 2. Primary commodity categories for Hampton Roads, Virginia. Commodity Category (Code) Average Annual Tonnage 2006–2010 (in 000s) @ 90% @ 100% Coal, Lignite & Coal Coke (10) 31,933 35,481 All Manufactured Equipment, Machinery and Products (70) 4,832 5,369 Sand, Gravel, Stone, Rock, Limestone, Soil, Dredged Material (43) 2,609 2,899 Other Agricultural Products; Food and Kindred Products (68) 1,932 2,147 Other Chemicals and Related Products (32) 1,762 1,958 Total 43,068 47,854 Total All Commodities 52,946 58,829

13 to dock-level data, the data had to be managed on the basis of river segments instead of on the basis of individual dock locations. Rather than having waterway segments/links with a single dock or facility, the model is required to treat whole river segments as links with a single landside connection, which means that multiple facilities are lumped together as one. This in turn requires some generalized assumptions about the facilities in each segment that may not be entirely accurate. Using more granular data, future research efforts can build on the proposed methodology to develop a more accurate evaluation of project alternatives. The data for calendar years 2006 to 2010 show that four commodity groups coded at the two-digit level make up over 81 percent of the total tonnage that originates, terminates, or passes through the Port of Huntington. An additional 21 groups make up the remaining 19 percent. Table 3 shows the top four commodity groups. Coal alone accounts for 64.4 percent of the total commod- ity tonnage flows for Huntington. Plaquemines, Louisiana (Dual Coastal/Inland) Coal and grain (corn and wheat) dominate the commod- ity mix for Plaquemines. The detailed data provided by the Corps reflected 90 percent of the actual total traffic. Table 4 shows the totals provided by the Corps and extrapolates them to the 100 percent level (internal and local traffic not included). The seven commodity groups shown in Table 4 make up 81 percent of the total cargo flow for Plaquemines. These commodities had significant domestic shipment volumes; therefore, each commodity has an import corridor, export corridor, domestic inbound corridor, and domestic outbound corridor. Table 4 does not include internal or coastwise shipments in the totals. Portland, Oregon (Coastal) Wheat exports dominate the commodity mix for Portland- Coastal. The seven commodity groups shown in Table 5 make up 82.5 percent of the total foreign trade cargo flow for Port- land. Table 5 does not include internal shipments in the totals. Portland, Oregon (Inland) As in the case of Huntington, because the research team could not access dock-level data, the data had to be man- aged on the basis of river segments instead of on the basis of individual dock locations. Rather than having waterway segments/links with a single dock or facility, the model is required to treat whole river segments as links with a single landside connection, which means that multiple facilities are lumped together as one. This in turn requires some gener- alized assumptions about the facilities in each segment that Commodity Category (Code) Average Annual Tonnage2006–2010 (in 000s) Coal, Lignite & Coal Coke (10) 46,386 Sand, Gravel, Stone, Rock, Limestone, Soil, Dredged Material (43) 4,599 Distillate, Residual & Other Fuel Oils; Lube Oil & Greases (23) 4,159 Gasoline, Jet Fuel, Kerosene (22) 3,354 Total 58,498 Total All Commodities 72,014 Table 3. Primary commodity categories for Huntington, West Virginia. Table 4. Primary commodity categories for Plaquemines, Louisiana. Commodity Category (Code) Average Annual Tonnage 2006–2010 (in 000s) @ 90% @ 100% Coal, Lignite & Coal Coke (10) 10,155 11,283 Corn (63) and Wheat (62) 4,959 5,510 Crude Petroleum (21) 3,002 3,336 Petroleum Pitches, Coke, Asphalt, Naphtha and Solvents (24) 2,431 2,701 Distillate, Residual & Other Fuel Oils; Lube Oil & Greases (23) 2,079 2,310 Oilseeds (Soybean, Flaxseed and Others) (65) 2,072 2,302 Gasoline, Jet Fuel, Kerosene (22) 1,180 1,311 Total 25,878 28,753 Total All Commodities 31,402 34,891

14 may not be entirely accurate. Using more granular data, future research efforts can build on the proposed methodology to develop a more accurate evaluation of project alternatives. The data for calendar years 2006 to 2010 show that four commodity groups coded at the two-digit level make up over 93 percent of the total inland waterway tonnage that origi- nates, terminates, or passes through the Port of Portland. (Forest products make up the remainder.) Table 6 shows the top four commodity groups. Multimodal Connections for the Selected High-Volume Freight MTS Corridors For the selected port areas, the researchers identified the landside corridors that tied into the waterfront origins and destinations for the primary commodity flows. Initially, the researchers intended to use the Corps’ confidential waterborne trip data file to identify specific points of interchange with the landside system (docks). However, the Corps determined that it could not make the data available to the researchers. This obstacle primarily affected the analysis of inland waterway ports. The data from CPT that were made available by the Corps were aggregated by river segment. Nearly all of these segments included multiple facilities, any one of which could have been the origin or destination of the waterborne flow(s) (these segments are referred to as links in the description of the model). This circumstance required the researchers to treat each segment as a link with one entry/exit point for landside connections in the modeling framework, even though a seg- ment might actually consist of multiple facilities. By examining the relevant waterfront facilities in each segment, the research- ers were able to make some assumptions as to whether the land- side corridor would be primarily highway, rail line, or pipeline. Because of the difficulties in assessing pipeline capacity and efficiency, and given that the primary focus of this research was the interconnection with rail and highway corridors, pipe- line flows were dropped from consideration. Given a railway or highway connection, further investigation was conducted to determine the most likely rail line(s) or highway(s) used to transport the commodity to or from the waterfront facility. The evaluation of waterfront facilities on inland waterways was conducted in a multi-phased approach. First, the research- ers examined the highest volume origin-destination pairs for the highest tonnage commodities originating, terminating, or passing through the port. All of the traffic associated with these segments (not just the high-volume origin-destination flows) was tallied, and the top-ranking river segments, those that accounted for at least 80 percent of the overall cargo flows, were selected. The researchers then used Google Earth satellite images and the Corps’ facility data to identify water- front facilities in each of the selected segments that were likely to handle the commodity in question. Once these facilities were identified, an attempt was made to develop an overall characterization of the segment, i.e., a determination was made as to whether the majority of cargo flows to/from the facilities in the segment would most likely be by truck, rail, or pipeline, or whether the segment might be an end point in Commodity Category (Code) Average Annual Tonnage 2006–2010 (in 000s) Wheat (62) Other Chemicals and Related Products (32) Fertilizers (31) Gasoline, Jet Fuel, Kerosene (22) Distillate, Residual & Other Fuel Oils; Lube Oil & Greases (23) All Manufactured Equipment, Machinery and Products (70) Primary Non-Ferrous Metal Products; Fabricated Metal Prod (54)/ Primary Iron and Steel Products (Ingots, Bars, Rods, etc.) (53) Total Total All Commodities 9,441 3,178 2,484 2,182 1,212 977 1,129 20,603 24,911 Table 5. Primary commodity categories for Portland-Coastal. Table 6. Primary commodity categories for Portland-Internal. Commodity Category (Code) Average Annual Tonnage2006–2010 (in 000s) Wheat (62) Gasoline, Jet Fuel, Kerosene (22) Sand, Gravel, Stone, Rock, Limestone, Soil, Dredged Material (43) Distillate, Residual & Other Fuel Oils; Lube Oil & Greases (23) Total Total All Commodities 4,444 2,640 2,404 1,976 11,464 12,312

15 the corridor. Using a variety of sources, the researchers then determined the most likely principal origins/destinations for the commodity and (in the case of non-pipeline flows) the most likely highway and rail lines used. Appendix C contains a description of the primary origin- destination corridors and the modal assignments for each corridor for each port. Only the ports selected for the case studies are included in the appendix, but a corridor analysis was done for each candidate port initially presented to the NCFRP Project 42 panel. Finally, some measure of system component availability for each corridor (whether on land or water) was needed to evalu- ate investment options. It was assumed that a project that would enhance resource availability and tap into a highly reliable net- work would be preferable to a project that would tap into a highly congested or unreliable network. The goal of the mea- sure was to define the current (without project or before condi- tion) utilization of the asset and the post-project (improved or after condition) utilization of the asset and compare the two. Initially, the researchers defined the measure as “capacity”; however, capacity was determined to be too vague—it could be affected by time of day, season of the year, equipment choices, and other variables. A term such as “utilization” focuses on the system itself, its historic activity, and its potential usefulness to a carrier/shipper. The details of how this measure was developed are described in the section of this report titled “Utilization Metrics/Indicators.” Method of Analysis Maintenance projects are not currently evaluated at a detailed level; rather, the objective is to maintain each naviga- tion project at its authorized dimensions. However, given that it is unlikely that budgeted amounts will make this objective achievable any time in the foreseeable future, a system must be developed to prioritize expenditures. There are a number of metrics that could be used to prioritize the possible O&M budget items. These include cargo value, cargo quantity, eco- nomic impact on the region, availability of alternative modes of transportation, and others. The research team chose to focus on the maximization of cargo flows based on tonnage. This measure could be expanded or even substituted in the model discussed in this report. Because there is currently no information available on how much an O&M project might affect cargo flows, the model is not able to maximize cargo throughput capability in an abso- lute sense; rather, it was necessary to employ a sensitivity-type analysis. The conceptual approach was to assume a certain loss of water depth due to lack of maintenance dredging. A loss of depth (draft) would not necessarily affect cargo flows in terms of total tonnage, although it would definitely affect the unit cost of the flow. Therefore, the assumption was made that loss of depth would not necessarily affect the volume of traffic. However, if the navigation project were to be restored to its authorized dimensions and remain there, it is highly probable that further shipments and investment in waterfront facilities would be made. The researchers analyzed what historically moved within affected stratum at each port and then analyzed a 10 percent, 20 percent, or 30 percent increase in that subset of the port’s traffic. Initial results showed that in many cases the 10 percent and 20 percent levels had insignificant effects. In fact, in some cases even a 30 percent increase in the tonnage at the affected depth level caused a very small increase in the port’s total traffic in some port communities. In order to simplify the analysis and make the results meaningful, the researchers chose to use only the 30 percent level for this research. Tonnage that is assigned to the affected stratum of the waterway is referred to as “project depth tonnage” in this research. Since this is the ton- nage that dredging will directly affect, the objective function in the model attempts to maximize this variable. The model recognizes the possibility that these potential increases could overtax the surface transportation system involved in the supply chain corridor. What follows is a step- by-step description of the method that was used to calcu- late the capacity of highways and railroads and the potential congestion resulting from additional tonnage generated by a potential maintenance project. Figure 1 shows a flow chart summarizing the general method of analysis. In several cases, much of this analytical work was not nec- essary. The preliminary corridor analysis and historical traffic patterns indicated that either rail or highway (or both) would not be affected. This is explained in the summary of the find- ings for each port. Step 1: Determine the Potential Tonnage Increase Determine the potential increase in utilization for each port as the maximum tonnage that could be potentially added to the port’s cargo volumes as a result of a project. For locks, the change in potential utilization (or theoretical capacity) is a function of the reduction in delay. While delays can be caused by unusual arrival patterns or weather conditions— and therefore cannot be eliminated—average delays are heav- ily influenced by down time and operational issues. For the purposes of this modeling exercise, it was assumed that delays could be cut by a certain amount with a maintenance proj- ect. For channel projects (i.e., in the case of both inland and coastal ports) potential utilization is essentially the number of additional tons that could be transported per foot dredged. As explained earlier, the projected increase in utilization is calculated by taking the tonnage that has historically transited in the depth stratum that the maintenance project will affect and then increasing that tonnage by 30 percent.

16 Step 2: Assign Tonnage to Each Corridor and Mode Assuming that a cargo increase will need to tap into a rail or highway corridor, the researchers used FAF3 data for 2011 to determine the percent distribution of potential tonnage by mode to/from the FAF3 area corresponding to the particular port and identify the major highway and rail corridors that transport it. Step 3: Determine How Much of the Potential Additional Tonnage Can Be Accommodated For the rail segments or highway segments in question, the current traffic levels and congestion levels were determined. The potential utilization of a corridor—defined as the increase in tonnage up to the point at which unacceptable congestion occurs—was then calculated. The method for determining N Convert to addional trucks on highway Convert to addional trains on rail Acceptable? Assignment to corridors and modes Addional tons on rail Addional tons on highway Calculate new level of congeson Calculate new level of congeson Acceptable? Use current plus acceptable new tonnage as tonnage level aer project Reduce tonnage to acceptable Addional tonnage throughput at port due to project (+10%, +20%, +30%) Use current plus acceptable new tonnage as tonnage level aer project Reduce tonnage to acceptable Sum for total project effect N Y Y Figure 1. Overview of method for determining capacity input values for model.

17 congestion and unacceptability is described in the modal sec- tions below. The model sets the potential tonnage increase for a corridor to the lesser of (1) potential total new tonnage or (2) total cargo flow at the point where congestion becomes unacceptable. (If the highway or rail mode has enough capacity to handle the projected cargo increase, it is essentially treated as unlimited capacity in the model.) Utilization Metrics/Indicators The approach used to develop utilization metrics/indicators depends on the type of system component: lock, river/channel, highway, or rail line. The following paragraphs describe the data the research team used and the corresponding methods that were employed to develop these metrics. Locks Capacity/Utilization In the case of lock capacity, the constraint is not depth. Lock capacity/utilization is greatly dependent on available operat- ing time and tow processing time. By improving the condition of the lock, it is possible to reduce the hours the lock is out of service and increase the capacity of the lock to move addi- tional project depth tonnage over a given time frame. The researchers conducted a thorough literature search for methodologies that have already been developed for determin- ing lock capacity on the inland waterway system. Only two were found that appeared to be methodologically sound. In the first case, the Corps did an exhaustive analysis of lock capacities on the Ohio River as part of its “Ohio River Mainstem System Study Integrated Main Report” of 2006 (8). The Corps performed a series of simulations for each lock using its proprietary Water- way Analysis Model (WAM). In the second case, the Oak Ridge National Laboratory (ORNL) did a mathematical/statistical analysis of a number of locks and developed a regression equa- tion that estimated the tonnage capacity of a lock based on historical tow sizes, tonnage throughput, and delay statistics. Appendix D contains a detailed description of this methodology. Since the ORNL analysis used publicly available data, the researchers used the ORNL approach to calculate the capacity of a number of locks on the Ohio River and then compared the results to what the Corps calculated using WAM. After factoring out obvious anomalies (outliers) in the data (typically caused by major accidents), the results of the two were acceptably simi- lar. Since the ORNL approach uses publicly available data and was acceptably close to the WAM results, the researchers calcu- lated capacities using the ORNL formulas. Table 7 shows the results for the Ohio River System locks included in the corridors identified in the case studies. Table 8 shows the Columbia/Snake River locks. Lock Practical Capacity (Tons) 2010 Tonnage % Utilization 2012 Tonnage % Utilization Peak Tonnage Peak Year Peak % Utilization New Cumb 120,714 26,289 22% 31,108 26% 35,252 2002 29% Pike 142,160 30,026 21% 31,751 22% 43,628 2002 31% Hannibal 166,201 42,284 25% 40,389 24% 53,288 2005 32% Willow 158,801 41,780 26% 39,786 25% 50,164 2005 32% Belleville 183,809 44,560 24% 42,165 23% 52,888 2005 29% Racine 151,951 45,611 30% 43,177 28% 52,275 2004 34% Byrd 183,998 50,398 27% 48,098 26% 62,481 2005 34% Greenup 140,903 56,443 40% 50,527 36% 71,709 2000 51% Meldahl 142,930 57,738 40% 51,618 36% 63,809 2001 45% Markland 156,442 57,595 37% 57,616 37% 61,401 2011 39% McAlpine 155,429 67,660 44% 71,105 46% 73,589 2011 47% Cannelton 152,647 67,974 45% 69,460 46% 72,305 2011 47% Newburgh 232,639 78,302 34% 78,934 34% 81,829 2011 35% Myers 184,744 71,501 39% 68,098 37% 75,279 2001 41% Smithland 321,261 78,405 24% 73,447 23% 85,915 2001 27% Monon L&D 2 37,406 14,832 40% 15,088 40% 21,733 2002 58% Emsworth 19,970 15,326 77% 16,520 83% 23,687 2002 119% Dashields 40,208 16,365 41% 17,897 45% 24,516 2002 61% Montgomery 34,688 18,237 53% 18,756 54% 26,709 2002 77% L&D 52 116,535 89,878 77% 91,401 78% 97,325 2005 84% L&D 53 230,592 79,628 35% 76,982 33% 89,153 2000 39% Shaded items are locks that could reach their capacity with increased tonnage from a proposed maintenance project. The corridor analysis considered these locks as potential choke points. Table 7. Lock utilization levels—Ohio River System (2000–2012).

18 The potential effect of a lock maintenance project was calcu- lated by identifying the lowest annual average delay from 2000 to 2012 and using it as the target (after project) condition. The difference between the theoretical capacity of the lock based on historical data and the theoretical capacity with the optimum delay factor would be the potential tonnage increase for the lock. As with highway and rail, the model compares the potential increase assigned to the corridor/lock as part of the port project analysis to the revised capacity of the lock, and the lesser of the two is taken as the actual effect of the maintenance project. One important assumption underlies the calculation of the indicator for locks—the size and composition of tows will not change as a result of the lock project under consideration. This means that the number of barges per tow will not change from the historical average and neither will the percentage split between loaded and unloaded barges. In other words, the average throughput per lockage will remain the same. Delays At locks, the effect of maintenance is to reduce down time (increase operating hours) and/or reduce delays. For this study, the focus is on a reduction in delay, which increases the poten- tial throughput at the lock. A unit of improvement is defined as one-third of the potential reduction in delays (the difference between historical average delay and optimum delay). In the case studies, up to three units of improvement are incorporated into a lock maintenance project, which amount to 100 percent of the potential reduction. For example, if the current aver- age delay is 0.90 hours and it could potentially be reduced to 0.30 hours, the difference is 0.60 hours. One-third of 0.60 hours is 0.20; this would be one unit of improvement. If three units of improvement are achieved, the new delay will be 0.30 hours. Channels For inland port waterway segments and coastal ports, the researchers used a metric based on channel depth as a uti- lization indicator. By examining historical average tonnage throughput at various draft levels during periods when ade- quate channel depth was available (as CPT does), it is possible to develop the tonnage by commodity that has historically transited in the depth stratum being affected by maintenance dredging. For the purposes of this study, it was assumed that each port would silt in 3 ft before maintenance would actu- ally be funded, with the exception of Huntington, where the assumption was 2 ft. It was assumed that the historical tonnages would have moved on the waterway even with reduced water depth; they would just have to move in smaller barge/vessel loads, and an additional number of trips would be required to con- tinue moving the same tonnage. The researchers made an assumption that if the waterway/channel were to be restored to its authorized dimensions and maintained there, it would be plausible to assume that additional tonnage would be attracted to the waterway. As noted earlier, this additional tonnage was termed “project depth tonnage.” A project that has silted in 3 ft (2 ft in the case of Huntington) would have zero capacity to move project depth tonnage. However, for each foot of dredging, a certain amount of proj- ect depth tonnage would be restored based on historical aver- ages, and a 30 percent increase in tonnage moving in the newly restored stratum was added on top of that for the purposes of this analysis. At the present time, there is no accepted methodology for determining the effect maintenance dredging (or the lack thereof) might have on tonnage throughput. It is for this rea- son that the Corps has selected historical tonnage and regional significance to prioritize maintenance dredging, rather than an actual analysis of potential impacts. Since there is no method- ology for determining what the tonnage increase from main- tenance dredging would be, the researchers initially employed a sensitivity analysis approach in which they increased the tonnage that historically moved in the affected stratum by 10, 20, and 30 percent to see how that would affect the model results. However, it became apparent early in the modeling Table 8. Lock utilization levels—Columbia/Snake River System (2000–2012). Lock Practical Capacity (Tons) 2010 Tonnage % Utilization 2012 Tonnage % Utilization Peak Tonnage Peak Year Peak % Utilization Bonneville 23,394 8,397 36% 8,670 37% 10,614 2000 45% The Dalles 13,225 8,025 61% 8,226 62% 10,141 2000 77% John Day 10,319 7,368 71% 7,259 70% 9,098 2000 88% McNary 21,652 6,244 29% 6,187 29% 8,409 2000 39% Ice Harbor 9,195 2,887 31% 3,175 35% 4,525 2000 49% Lower Monumental 8,232 2,554 31% 2,776 34% 4,090 2000 50% Little Goose 8,233 2,233 27% 2,593 31% 3,059 2000 37% Lower Granite 3,296 1,265 38% 1,510 46% 2,183 2000 66% Shaded items are locks that could reach their capacity with increased tonnage from a proposed maintenance project. The corridor analysis considered these locks as potential choke points.

19 process that in many cases, even a 30 percent increase in the potentially affected tonnage stratum was not a significant number in relation to the total tonnage each port handles. The researchers used 30 percent as the expected increase for purposes of testing the model and its indicators. As an illustration, assume that a port has 100 million tons of traffic each year. Also, assume that 70 million tons of that traffic is project depth tonnage; it needs the maximum depth. The expected increase of 30 percent is applied to the 70 mil- lion tons, resulting in a potential increase of 21 million tons. Note that a 30 percent increase in project depth tonnage is a 21 percent increase in the port’s total tonnage. The analysis treated each inland waterway segment that is part of an origin-destination pairing as a non-lock node—i.e., as if it were a channel of a coastal port—for purposes of determining highway and rail capacities. In other words, the selected river segments were potentially origin or destination points for a surface transportation corridor as well as origins or destinations for waterborne shipments. In the calculation of how much tonnage could be potentially added to particu- lar waterway segments associated with each such node in the case of an inland waterway port, constraints caused by water depth and lock throughput capacity were both considered. The current theoretical availability and the projected increase resulting from any project were expressed in terms of ton- nage, as explained above. For inland port analysis (Huntington and Portland- Inland) the analysts recorded all segments and locks included in the identified corridors as well as their sequence. For each segment or lock that was analyzed, the potential additional tonnage to be gained from the project was expressed in tons per foot dredged (for segments) or tons per unit of reduction in delays (for locks). Highways The highway corridor analysis tapped into previous and ongoing work conducted by TTI through its Urban Mobility Research Program (UMRP) (9). Based on its previous work across the country, UMRP was able to determine that any potential highway congestion effects from increased port traffic would only be noticeable in the immediate vicinity of the port. The researchers established how many more trucks could be added to the arterials that link ports to Interstate or U.S. highways before congestion levels began to show dra- matic deterioration. In order to calculate the number of truck trips that would be generated by new tonnage throughput resulting from a proj- ect, the researchers assumed an average truckload of 25 tons per truck. They did not attempt to factor in backhauls, since there was no way to determine whether the truck would return directly to its origin, and if it did, what route it would follow. Rail The capacity analysis of potential rail corridors consisted of the following steps: 1. Define relevant rail corridors for analysis. 2. For each corridor, identify terminals and/or rail line seg- ments where congestion is likely to be severe. 3. Determine the theoretical train count capacity of the ter- minal or rail line segment by using a simplified approach identified in the National Rail Freight Infrastructure Capacity and Investment Study prepared for the Association of American Railroads by Cambridge Systematics, Inc., in 2007 (2). 4. Establish current train counts for the identified terminal or rail line segment using data from the U.S. DOT National Highway-Rail Crossing Inventory Program. 5. Calculate the remaining available capacity by subtracting (4) from (3) above. 6. Using a standard train size and carload tonnage (bearing in mind the primary commodities being transported), cal- culate the equivalent tons of remaining available capacity. Corridor Analysis The tonnage for each primary commodity-corridor com- bination was divided among the highway, rail, or water modes, based on the FAF3 data. Appendix C provides the modal assignment for the corridors associated with each port. Once the corridor modes and tonnages were established, the researchers assessed each corridor to see if there were any constraints that would prevent the tonnage increases from flowing through the corridor. Appendix E discusses the con- straints. The manner in which this information was input into the model is described next.