National Academies Press: OpenBook

Design of Construction Work Zones on High-Speed Highways (2007)

Chapter: Chapter 4 Methodology and Findings

« Previous: Chapter 3 Current Work Zone Design Guidance
Page 51
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 51
Page 52
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 52
Page 53
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 53
Page 54
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 54
Page 55
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 55
Page 56
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 56
Page 57
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 57
Page 58
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 58
Page 59
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 59
Page 60
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 60
Page 61
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 61
Page 62
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 62
Page 63
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 63
Page 64
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 64
Page 65
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 65
Page 66
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 66
Page 67
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 67
Page 68
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 68
Page 69
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 69
Page 70
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 70
Page 71
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 71
Page 72
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 72
Page 73
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 73
Page 74
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 74
Page 75
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 75
Page 76
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 76
Page 77
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 77
Page 78
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 78
Page 79
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 79
Page 80
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 80
Page 81
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 81
Page 82
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 82
Page 83
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 83
Page 84
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 84
Page 85
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 85
Page 86
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 86
Page 87
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 87
Page 88
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 88
Page 89
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 89
Page 90
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 90
Page 91
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 91
Page 92
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 92
Page 93
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 93
Page 94
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 94
Page 95
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 95
Page 96
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 96
Page 97
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 97
Page 98
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 98
Page 99
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 99
Page 100
Suggested Citation:"Chapter 4 Methodology and Findings." National Academies of Sciences, Engineering, and Medicine. 2007. Design of Construction Work Zones on High-Speed Highways. Washington, DC: The National Academies Press. doi: 10.17226/14032.
×
Page 100

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.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 51 CHAPTER 4 METHODOLOGY AND FINDINGS 4.1 DEVELOPMENT OF ROADSIDE DESIGN AND TEMPORARY BARRIER PLACEMENT GUIDANCE FOR CONSTRUCTION WORK ZONES Task 3 of NCHRP Project 3-69 called for “a survey of the states to collect guidance related to construction work zones and traffic control.” At the time that the survey was created, it appeared that previous research and the associated literature would not provide direction as to which design features are high risk factors in work zones and should be prioritized for selected research during phase II of the project. This indeed was the case (see Chapter 2 of this report). Therefore, an additional objective of the survey became: determine priority topics associated with the design of construction work zones using state DOT input. Of the 32 states that responded to the survey, 24 ranked having or improving guidance on traffic barriers and roadside design as “most important/critical.” This was the highest ranking topic of those included in the survey or added by DOT responses. Therefore, a specific study to address this issue was proposed in the first interim report, and the NCHRP project panel approved the study for execution during phase II of this project. This section documents the methodology and results of that effort. In addition, the results are also incorporated into Chapter 5 of the design guidance (Appendix A), Roadside Design and Barrier Placement. 4.1.1 Research Methodology Information for completion of this study and development of guidance would come from four sources: • Roadside principles and practices for permanent roadways; • Completed and ongoing research related to roadside safety in construction work zones; • State DOT roadside design guidance for construction work zones; • Incremental benefit-cost analysis of work zone scenarios. Each of these sources and their pertinence to this study are reviewed below. Section 4.1.6, Integration and Fusion to Develop Roadside Design Guidance, discusses the background and source of all relevant information used to develop practical guidance for roadside design and placement of temporary traffic barriers for construction work zones on high-speed highways.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 52 4.1.2 Roadside Principles and Practices for Permanent Roadways Adoption of the roadside safety principles and the implementing procedures outlined in the Roadside Design Guide has significantly enhanced highway safety. The forgiving roadside concept, clear zone, prioritized treatment of hazards, and crashworthiness are applicable to work zones as well as permanent roads. The forgiving roadside concept is based on the premise that “Regardless of the reason for a vehicle leaving the roadway, a roadside environment free of fixed objects with stable, flattened slopes enhances the opportunity for reducing crash severity.” An integral part of the forgiving roadside concept was the establishment of a clear zone, a traversable and unobstructed roadside area. When this concept was introduced in the AASHTO Yellow Book (40) in 1974, the dimension of the desired clear zone, 30 feet, was based on studies that indicated that 80 percent of the vehicles leaving the roadway could recover within this distance. Because of the perceived impracticality and sometimes inadequacy of this dimension for variable volumes and speeds and for different roadside slopes, variable desired clear zones were introduced in 1977 by AASHTO’s Guide for Selecting, Locating and Designing Traffic Barriers (41). Variable clear zones, based on design speed, volume, and roadside slopes, with adjustments based on the horizontal alignment, can still be found in the 2002 edition of the Roadside Design Guide. Where objects are located in the roadside, and especially within the desired clear zone, a series of alternative actions should be considered to reduce the risk to errant vehicles. The order of preference for addressing roadside obstacles follows: 1. Remove the obstacle; 2. Redesign the obstacle so it can be safely traversed; 3. Relocate the obstacle to a point where it is less likely to be struck; 4. Reduce impact severity by using an appropriate breakaway device; 5. Shield the obstacle with a longitudinal barrier designed for redirection, or use a crash cushion; 6. Delineate the obstacle if the above alternatives are not appropriate. Alternatives 4 and 5 introduce the concept of crashworthiness. Where conditions require the presence of an obstacle or barrier near the traveled way, it should be designed to perform appropriately (i.e., minimize probable motorist harm) if struck. Signs, signals, luminaire supports, and utility poles should be breakaway devices. Guidance for these objects is contained in AASHTO’s Standard Specifications for Structural Supports for Highway Signs, Luminaires, and Traffic Signals (42), A Policy on the Accommodation of Utilities within Highway Right-of-Way (43), and A Policy on the Accommodation of Utilities within Freeway Right-of-Way (44). Roadside barriers are deemed crashworthy by passing the crash test criteria of NCHRP Report 350, “Recommended Procedures for Safety Performance Evaluation of Highway Features” (45).

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 53 The clear zone concept has been widely accepted because of its perceived simplicity. In general, the idea has been to observe the clear zone of a roadway segment. If there are objects that present a potential safety hazard to a motorist when struck, analyze and treat the hazard with one of the prioritized treatments. However, the current edition of the Roadside Design Guide gives the following instruction: A basic understanding of the clear zone concept is critical to its proper application. The numbers obtained…imply a degree of accuracy that does not exist… In some cases, it is reasonable to leave a fixed object within the clear zone; in other instances, an object beyond the clear zone distance may require removal or shielding. Similar discussion complements the above instruction: …to include every recommendation or design value in this chapter (Chapter 3) on every future highway project is neither feasible nor possible. Engineering judgment will have to play a part in determining the extent to which improvements can reasonably be made with the limited resources available. These excerpts indicate that application of the clear zone approach often involves subjective roadside safety decisions. The use of incremental benefit-cost analysis to roadside conditions is one means of reducing the level of subjective judgment. 4.1.2.1 Benefit-Cost Analysis for Permanent Roadways Benefit-cost analysis is “a method by which the estimated benefits to be derived from a specific course of action are compared to the costs of implementing that action (6).” The benefits usually refer to reduced crash or societal costs as a result of decreases in the number and/or severity of crashes. The costs of implementing the action are the direct costs to the highway agency for initial installation, maintenance, and repair costs. If the ratio of benefits to costs (equation 3) exceeds 1, then the benefits derived will be equal to the investment over the analysis period. The benefit-cost ratio can be used to compare several different actions against each other and against the no action alternative. B/C Ratioj-i = ij ji DCDC CCCC − − (3) Where B/C Ratioj-i = Incremental benefit-cost ratio of alternative j to alternative i; CCi, CCj = Crash or societal costs resulting from crashes under alternatives i and j (annualized over the analysis period); DCi, DCj = Direct costs for alternatives 1 and 2 (annualized over the analysis period).

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 54 A benefit-cost ratio greater than 1 does not alone justify the implementation of a particular alternative. However, observing the ratios provides designers or other decision makers with quantitative information to help in making the best investment for safety and mobility needs. To perform a benefit-cost analysis based on safety, several tools need to be available for the analyst: • Method to predict crash frequencies under all proposed alternatives. • Method to predict crash severities under all proposed alternatives. • Crash cost estimates by severity; • Repair cost estimates; • Installation and maintenance costs for specified safety treatments; • Discount rate over the analysis period. Expected crash frequencies are difficult to predict because of the infrequency and randomness of the event. Different methods to do so include (1) crash prediction models, which are usually regression models used to predict crashes given the roadway geometry, segment length, and traffic, (2) historical data on the roadway of interest or similar roadways, and (3) simulation. The last is the most common method for predicting roadside crashes. The simulations are usually based on an encroachment model that predicts the frequency of encroachments as well as the encroachment speed, angle, and lateral extent of the encroachment. Knowing these variables as well as the layout of the roadside, it can be predicted whether or not a crash would occur. The weaknesses of these simulations are that the results are only as good as the underlying encroachment models, the state of which has not been advanced much since the 1970s. Crash severities can also be predicted in several ways. As in predicting frequencies, these include (1) logistic regression, used to predict crash severity given a series of predictor variables including roadway geometry, (2) historical data on the roadway of interest or similar roadways, and (3) simulation. As with predicting frequencies, the accuracy of simulating crash severities is dependent on the reliability of the underlying algorithms. These algorithms are sometimes based on historical data of crashes with different objects for a range of impact conditions. However, data of this detail for a range of objects are not widely available, and experience and judgment are sometimes used. Crash costs for varying crash severity levels are calculated by estimating the results of a motor vehicle crash and the effects of a specified injury on the involved persons’ entire lives. The most useful measure of crash cost is a comprehensive cost, which includes 11 different cost components: property damage; lost earnings; lost household production; medical costs; emergency services; travel delay; vocational rehabilitation; workplace costs; administrative, legal, and pain costs; and lost quality of life. Estimates of crash costs are usually published by several public and private organizations, all using different bases and assumptions. Therefore, a range of cost estimates exist.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 55 Repair costs consider the cost of repair of a safety treatment or other roadside object that has functional value after a crash has occurred with that object. Repair costs can be estimated from historical data, full scale crash testing, or simulation. For example, crash testing and simulation can be used to determine the length of damage to a guardrail or other safety treatment given a certain vehicle size and impact speed. The repair cost would then be the product of the length (or other unit) of the damaged safety treatment and the unit cost for repair. Installation and maintenance costs can usually be determined by a state DOT through historical records. For example, some states publish this type of price information based on contractors’ bid prices for standard bid items. Finally, discount rates are interest rates used to determine the current value of costs that will be incurred over the entire period of a benefit-cost analysis. It is the current value of the benefits and costs that are used in equation 3. Discount rates can be determined by observing the interest rate charged to commercial banks and other depository institutions on loans they receive from their regional Federal Reserve Bank's lending facility. Observational before-after studies are a more controlled type of study to determine the benefits of a safety treatment; however, the treatment must be applied and several years of before-and-after data accumulated before the benefit is determined. The planned Highway Safety Manual will include procedures to help state DOTs determine the benefits of safety countermeasures using these types of procedures. After years of implementation, historical data as a result of these analyses will exist to assist DOTs in predicting the benefits of future countermeasure applications. For a variety of reasons, conducting observational before-after studies to determine the effects of different safety treatments is often not viable. Therefore, the techniques used as part of benefit-cost analysis discussed above (e.g., crash frequency/severity prediction) will need to be continually updated and refined as data become available. The adaptation and application of the benefit-cost analysis procedures discussed above to construction work zones is discussed in Section 4.1.5. 4.1.2.2 Use of Roadside and Median Barriers for Permanent Roadways A roadside barrier “is a longitudinal barrier used to shield motorists from natural or man-made obstacles located on either side of a traveled way. It may also be used to protect bystanders, pedestrians, and bicyclists from vehicular traffic under special conditions (6).” Similarly, median barriers are longitudinal barriers used to separate opposing traffic on a divided highway. In either case, the purpose of a longitudinal barrier is to contain or redirect a vehicle that leaves the roadway and strikes it, with less severe consequences than if the barrier had not been there. Both roadside and median barriers must meet the performance criteria set forth in NCHRP Report 350, “Recommended Procedures for the Safety Performance Evaluation of Highway Features.”

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 56 Roadside and median barriers should only be installed where crashes with the barrier are likely to be less severe than crashes without the barrier. In addition, the roadside hazard being shielded should be exposed to a significant level of traffic over the performance period to justify the cost of providing and maintaining the barrier. Subjective analysis or benefit-cost procedures are the two methods discussed in the Roadside Design Guide for making these determinations. Roadside barriers are used to shield errant vehicles from two basic categories of roadside conditions: embankments and roadside obstacles. Embankment height and side slope are the basic factors considered in determining potential harm to errant vehicles. Roadside Design Guide Figure 5.1 was developed based on the relative severity of encroachments on embankments versus impacts with roadside barriers. Figure 5.1 does not take into account the probability of an encroachment or the cost of leaving the slope unshielded versus the cost of barrier installation, maintenance, and repair. Therefore, from a benefit-cost standpoint, the figure most likely overestimates the need for barrier for lower-volume roads. Figures 5.2 and 5.3 of the Roadside Design Guide are modifications of the criteria in Figure 5.1 that do consider these additional factors. The charts are not included for application, but states are encouraged to develop similar criteria based on their own evaluations. An example is shown in Figure 10 below. Median barriers are used to separate opposing traffic on divided highways, through traffic from local traffic, or high occupancy vehicle (HOV) lanes from general purpose lanes. Median barriers are similar to roadside barriers except that they are designed to redirect vehicles striking either side of the barrier. Figure 6.1 of the Roadside Design Guide (Figure 11 below) provides suggested guidelines for median barriers on high-speed, controlled access roadways that have relatively flat, traversable medians. The criteria are based on a limited analysis of median crossover crashes and should be used in the absence of more current or site-specific data. For ADTs above 20,000 vehicles per day, Figure 11 suggests that the use of median barrier would provide some benefit. More recent studies may suggest that benefits exist for median widths of 70 feet or less. When a roadside hazard is present in a construction work zone, a decision must be made on whether it would be cost effective to shield the hazard. A temporary concrete barrier is the option most preferred by state transportation agencies for this purpose. One objective of this research was to develop design aids for commonly occurring construction work zone scenarios that are easy to use, similar to Figures 5.1 and 6.1 of the Roadside Design Guide. The results of a series of benefit-cost analyses would be the primary criteria for development of the design aids. The scenarios and resulting design aids are presented in Chapter 5 of Appendix A.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 57 Figure 10. Example design chart for cost-effective embankment warrants based on traffic speeds and volumes, slope geometry, and length of slope (Figure 5.3b from Roadside Design Guide) (6). 4.1.3 Existing and Ongoing Research on Construction Work Zone Roadside Design and Safety The two most relevant studies to the development of roadside design and barrier placement guidance were conducted by Sicking and Ross (46) and Michie (47). Sicking and Ross used a benefit-cost procedure to assess the need for positive traffic barriers in work zones. The procedure was applied to develop general guidelines for four typical activities: bridge widening, roadway widening, major structural work near a traveled way, and two-lane, two-way operation on a normally divided highway. The benefit-cost procedure was also used to evaluate end treatments for barriers, including flaring the barrier away from the traveled way and the use of crash cushions.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 58 Figure 11. Suggested guidelines for median barriers on high-speed roadways (Figure 6.1 from Roadside Design Guide) (6). Michie also used benefit-cost procedures to define typical construction zone activities where positive barriers are needed. His methodology utilized the AASHTO ROADSIDE computer program (available with the 1996 Roadside Design Guide) to generate estimates for the number of collisions in a work zone with and without placement of positive barrier. Michie based the severity indices of work-zone-specific features (i.e., equipment, workers) on the work of Sicking and Ross. The result of Michie’s work was a series of design charts for different traffic speeds (43.5 mph to 68.3 mph) and hazard types (edge drop-off, structures, workers, heavy equipment, light equipment). Given the number, length, and offset of a hazard, Michie’s charts will provide the threshold effective traffic volume (ETV) at the construction site required to justify temporary concrete barrier. The primary advantage of Michie’s charts is that a variety of hazard combinations can be analyzed for cost-effectiveness of shielding with a barrier. The primary disadvantage is the complexity of use when compared to design charts in the Roadside Design Guide.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 59 The research reported in Section 4.1.5 attempts to build on the work of Sicking and Ross, and Michie. Primary differences were (1) the use of the Roadside Safety Analysis Program (48), which is the most recent computerized procedure that performs cost-effective analyses of roadside safety treatments and (2) the final form of the design guidance in Appendix A of this report, which illustrate four different benefit-cost regions given the input parameters of exposure and speed. 4.1.4 Existing State DOT Construction Work Zone Roadside Design Guidance A review of state DOT guidance on roadside design and traffic barrier placement is provided in Chapter 3, section 3.2.5. In this section, gaps and needs in state DOT roadside design practices are identified. Additionally, current practice related to application of the clear zone concept in work zones is summarized. State DOTs that completed this research project’s Task 3 survey indicated that improved guidance on traffic barriers and roadside design was “most important/critical” and should be prioritized. Although results from the studies discussed in section 4.1.3 of this chapter have been published and available for some time, they have not been incorporated into work zone practice for determining barrier need. Instead, there is still considerable reliance on designer judgment and experience. Although judgment and experience will always be important factors in the provision of temporary concrete barrier to shield different work zone roadside hazards, this research should aim to fill the gaps in current state DOT guidance and provide more quantitative design tools. As discussed in section 4.1.2 of this chapter, although the clear zone concept has been prevalent in roadside design since the 1960s, it does not clearly provide a solution to whether it would be cost effective to treat a roadside hazard with one of the prioritized safety treatments. However, the clear zone concept has been widely accepted because of its perceived simplicity (e.g., if an object is within the clear zone, provide one of the prioritized treatments; if it is outside the clear zone, it does not present a significant hazard to drivers). Because of its acceptance, designers may wish to use the clear zone concept instead of, or in combination with, the benefit-cost procedure discussed in section 4.1.5 below. Therefore, the state of practice for work zone clear zones was reviewed, and representative dimensions were provided in the design guidance. The representative guidance was based on that of Illinois DOT. 4.1.5 Incremental Benefit-Cost Analysis for Work Zone Scenarios A major objective of this research was to develop easy-to-use barrier placement design aids for commonly occurring construction work zone scenarios. The results of a series of benefit-cost analyses would be the primary basis for development of the design aids. The design aids based on benefit-cost analysis are combined with information from various other sources to provide roadside design and temporary barrier placement guidance for construction work zones on high-speed highways.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 60 The Roadside Safety Analysis Program (RSAP) was considered by the research team to be the best available tool for developing work zone barrier placement guidance based on benefit-cost analysis. Available RSAP documentation does not include work zone analysis as a potential application of the program and inherent differences between work zones and permanent roadway situations were recognized. Selective departures from the RSAP default procedures were made to more closely represent the distinctive characteristics of work zones. A detailed review of these departures is provided in section 4.1.5.2. Section 4.1.5.1 provides a brief overview of the general RSAP algorithm. For a more detailed discussion, see NCHRP Report 492, “Roadside Safety Analysis Program (RSAP) – Engineer’s Manual” (48). 4.1.5.1 RSAP As discussed in section 4.1.2.1, to perform a benefit-cost analysis based on safety, several tools need to be available to the analyst: • Method to predict crash frequencies under all proposed alternatives; • Method to predict crash severities under all proposed alternatives; • Crash cost estimates by severity; • Repair cost estimates; • Installation and maintenance costs for specified safety treatments; • Discount rate over the analysis period. The objective of NCHRP Project 22-9, Improved Procedures for Cost- Effectiveness Analysis of Roadside Safety Features, was to develop a computerized cost- effectiveness analysis procedure that (1) would incorporate these tools into a program that was capable of assessing roadside safety improvements at spot locations over sections of roadway and (2) could be used for development of warrants and guidelines of safety features with different performance levels. The product of this research was RSAP. The following discussion briefly summarizes the RSAP algorithm. It is not meant to supersede any of the detailed descriptions and information in the project report (48). Crash Frequency. To predict crashes, RSAP uses an encroachment based model with a hazard envelope described in (49). The assumption behind estimating roadside crashes from the number of encroachments is that crash frequency is proportional to encroachment frequency. The link occurs by simulating encroachments, then (1) determining whether each encroachment is inside or outside the hazard envelope of a roadside object and (2) determining if the lateral extent of the encroachment is greater than or less than the lateral offset of the object. To determine a base encroachment frequency, RSAP uses data collected in the 1970s (50). The data were based on observations of tire tracks in the median and roadside. Therefore, vehicles that encroached onto a concrete shoulder were not represented. To account for this under-representation, the frequencies are adjusted by different factors for undivided and divided highways. Adjustments are also made to account for controlled encroachments, the presence of horizontal curves, and the presence

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 61 of vertical grades. The user can also input a User Defined Adjustment Factor to account for unusual situations that could affect encroachment frequencies beyond the parameters incorporated into the program. The path of the encroaching vehicle is a function of encroachment angle, vehicle size, and vehicle orientation (49). A straight path with no steering or braking is assumed in the current version of RSAP. The lateral extent of each encroachment is determined through cumulative distribution functions of lateral extent for undivided and divided highways. These functions were developed from re-analysis of the data in (50). The vehicle path is checked against the coordinates of the roadside features to determine if a crash would occur. If the encroachment would result in a crash, then the impact conditions are estimated. This includes speed, angle, and vehicle orientation. For each predicted impact with a roadside safety device such as a barrier or crash cushion, RSAP will check for penetration of the feature and subsequent impacts. Speed adjustments are made after each penetration, and if multiple crashes occur, the most severe will be used to calculate crash costs. Crash Severity. After a crash is predicted to occur, RSAP estimates the severity of the impact. Crash severity estimation is perhaps the most important and most difficult step in the cost effectiveness analysis procedure. The initial intent of the RSAP developers was to use a new methodology for estimating crash severity that would involve a combination of police level crash data and kinematics analysis. However, development was too extensive and time consuming to be completed under the project. Instead, the severity indices listed in the 1996 Roadside Design Guide are used in the current version of RSAP. The severity index (SI) scale is associated with fixed levels or percentages of fatality, injury, and property damage only (PDO) as shown in Table 8. Severity indices are intended to be representative of average crashes and are usually developed through engineering judgment and expert opinion. Before implementation into RSAP, modifications were made to the severity indices in the 1996 Roadside Design Guide. These included relating SI to impact speed rather than design speed and developing a linear regression line to relate SI and impact speed for different object types.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 62 Table 8 Relationship of severity index to crash severity (48) Injury Level (%) SI None PDO1 PDO2 C B A K 0 100.0 -- -- -- -- -- -- 0.5 -- 100.0 -- -- -- -- -- 1 -- 66.7 23.7 7.3 2.3 -- -- 2 -- -- 71.0 22.0 7.0 -- -- 3 -- -- 43.0 34.0 21.0 1.0 1.0 4 -- -- 30.0 30.0 32.0 5.0 3.0 5 -- -- 15.0 22.0 45.0 10.0 8.0 6 -- -- 7.0 16.0 39.0 20.0 18.0 7 -- -- 2.0 10.0 28.0 30.0 30.0 8 -- -- -- 4.0 19.0 27.0 50.0 9 -- -- -- -- 7.0 18.0 75.0 10 -- -- -- -- -- -- 100.0 C = minor or possible injury. B = moderate or non-incapacitating injury. A = severe or incapacitating injury. K = fatal injury. Crash Cost. After the severity index of a crash is estimated, the crash or societal costs associated with the crash are calculated by multiplying the probability of each level of injury by the cost associated with that injury. RSAP provides the alternatives of four different sets of crash cost figures that can be used for the analysis: • Cost figures from the Roadside Design Guide; • FHWA comprehensive crash cost figures; • User-defined crash cost figures categorized as fatal, severe injury, moderate injury, minor injury, and PDO; • User-defined crash cost figures categorized as fatal, injury, and PDO. Because of a scaling procedure used in RSAP to ensure that low probability, high cost events have adequate representation in a series of runs, the initial crash costs must be adjusted by a weighting factor. This procedure is discussed in Chapters 4 and 8 of (48). Repair Cost. The cost of repairing roadside safety hardware in RSAP is determined by correlating repair costs to impact energy terms. Depending on the impact conditions, the amount (e.g., length) of a roadside safety hardware device that would need repair is determined and multiplied by the unit cost for the repair. Because the repair costs are based on probabilistic events (i.e., the impact conditions), they are weighted in the same manner as the crash costs. Repair costs represent average repair costs and can

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 63 often introduce inaccuracies. However, these inaccuracies in repair costs are insignificant when compared to crash and installation costs and do not affect the overall benefit-cost analysis. Installation Costs, Maintenance Costs, and Discount Rate. Installation and maintenance costs for each safety treatment in an alternative are user inputs. Installation cost is entered as a lump sum cost, and maintenance costs are entered as an annual value. This information can usually be determined by a state DOT through historical records. For example, some states publish this type of price information based on contractors’ bid prices for standard bid items. Finally, a discount rate can be entered by the user that represents the real cost of borrowing money, measured by the difference between interest rates and annual inflation rate. RSAP uses a 4 percent discount rate as a default value; however, a different rate may be used if deemed appropriate by the analyst. 4.1.5.2 Adaptation of RSAP for Work Zone Analysis The intended use of RSAP is to evaluate roadside safety alternatives for permanent roadways. Compared to permanent roadside situations, there are many factors that are inherently different in construction work zones; the most important of these factors are the number and types of safety hazards and the level of exposure to particular hazards. For example, barrier placement decisions for permanent roadside hazards may be evaluated over a 25-year analysis period, whereas construction work zone hazards may exist only for a few days to 24 months. In addition, common construction work zone hazards (e.g., clusters of equipment) are not normally considered for permanent roadway analysis. To address the difference in exposure, each work zone scenario was run with a one-year analysis period (the minimum possible with RSAP), then multiplied by the ratio of the work zone duration to one year. For example, if the work zone scenario being analyzed lasted four months, then the one-year benefit-cost ratio would be multiplied by 0.333. An advantage for this approach was that for a one-year run, the benefit-cost ratio for an infinite number of exposure levels could be determined. This calculation is simplified in that it does not multiply the repair costs (in the denominator) by the respective duration. However, since the repair costs were small compared to crash and installation costs, the benefit cost ratios were not affected. To address the different features that would be important for work zone analysis, but were not available in the RSAP object menus, the research team attempted to use the User Defined Feature capability of RSAP. However, it was discovered and later confirmed through communications with the RSAP development team that this capability was not operating correctly in the most recent version of the program. As an alternative, the characteristics of objects available on the RSAP object menus (but not otherwise used) were manipulated and used as surrogates. This approach was successful. RSAP includes a series of files with #.dat extensions, which store information about different roadside features, including name, severity index at a 0 mph impact speed,

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 64 change in severity index with impact speed, and average repair cost per crash. This information was modified in the si5.dat file for fixed objects. Four types of breakaway sign supports, not otherwise used for the work zone analysis, were changed to opposing vehicles, workers, heavy equipment, and light equipment (see Figure 12). Attempts to model head-on collisions with opposing vehicles using RSAP were unsuccessful. The latter three objects were common to most work zone scenarios and will be discussed here. Figure 12. Manipulated RSAP fixed objects pull-down menu. The worker object was defined as an area occupied by workers. Heavy equipment represented construction zone hazards that are rigid and heavy and that would not deflect or move from impact by a motor vehicle (e.g., cranes, paving machines, milling machines, compactors). Light equipment includes less massive items such as welding machines, compressors and pick-up trucks. The severities of impacts associated with these objects as well as the average repair costs per crash are summarized in Figures 13, 14, and 15.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 65 Severity Index vs. Impact Speed (Workers) 0 1 2 3 4 5 6 7 8 9 10 0 10 20 30 40 50 60 70 Impact Speed (mph) Se ve rit y In de x SI = 10 Average repair cost = $2000 Figure 13. Severity index and repair costs for motor vehicle-worker crashes. Severity Index vs. Impact Speed (Heavy Equipment) 0 1 2 3 4 5 6 7 8 9 10 0 10 20 30 40 50 60 70 Impact Speed (mph) Se ve rit y In de x SI = 0.1429 * impact speed Average repair cost = $8000 Figure 14. Severity index and repair costs for motor vehicle-heavy equipment crashes.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 66 Severity Index vs. Impact Speed (Light Equipment) 0 1 2 3 4 5 6 7 8 9 10 0 10 20 30 40 50 60 70 Impact Speed (mph) Se ve rit y In de x SI = 0.1143 * impact speed Average repair cost = $4000 Figure 15. Severity index and repair costs for motor vehicle-light equipment crashes. The RSAP base encroachment rate was also adjusted to reflect differences between permanent roads and work zones. As discussed earlier, the RSAP encroachment frequency was developed using data collected in the 1970s (50) and with several adjustments. A User Defined Adjustment Factor is available to account for factors affecting encroachment frequencies beyond the parameters incorporated into the program. In general, previous studies have shown that crash risk in work zones is higher than on permanent roadways. Increases in the number of crashes in a work zone compared to pre-work zone conditions generally ranged from 7 to 99 percent (see section 2.3.1). The results of several studies showed a crash increase between 20 and 40 percent. In addition, one study showed that work zones with lane widths less than 12 feet experienced a higher number of crashes than work zones with 12-foot lane widths (see section 2.2.4). Given those findings, the assumption that crashes are proportional to encroachments, and that RSAP makes no adjustment to encroachment rates depending on lane width, the following user-defined encroachment adjustment factors were used: • 1.4 for 12-foot lane widths in work zones; • 1.5 for 11-foot lane widths in work zones; • 1.6 for 10-foot lane widths in work zones.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 67 The other assumptions inherent in all scenarios were the crash costs and direct costs for the safety treatments. FHWA KABCO crash costs from technical advisory T 7570.2, Motor Vehicle Accident Costs (51), were escalated to 2004 values using the Gross Domestic Product (GDP) implicit price deflator. The final values are shown below: • Fatal, $2,938,000; • Severe injury, $203,400; • Moderate injury, $40,680; • Minor injury, $21,470; • PDO, $2,260. A primary source for the safety treatment costs was Pennsylvania Department of Transportation’s Publication 287, Construction Costs Catalog for Standard Construction Items (52). When the information was available, results of internet searches were also used to find a nationwide representative cost. Costs that were used for the commonly modeled items were: • TL-3 portable concrete barrier, $27 per linear foot; • TL-3 strong post guardrail, $9.00 per linear foot; • TL-3 temporary impact attenuating device, $4000 each; • Guardrail end treatment, $600 each. For each common work zone scenario, a series of RSAP runs were made using various combinations of average daily traffic, project duration and posted speed. Average daily traffic directly influences encroachment rate and when combined with project duration influences encroachment frequency. Posted speed influences encroachment speeds, impact speeds and crash severity. Therefore, given a work zone roadside scenario, benefit cost ratios will change as these three variables (average daily traffic, project duration, posted speed) change. In general, the following relationships can be expected: • As average daily traffic increases, the encroachment rate and benefit-cost ratio will increase; • As project duration increases, the encroachment frequency and benefit-cost ratio will increase; • As posted speed increases, crash severity and benefit-cost ratio will increase. The results of the RSAP runs were plotted in two-dimensional space capturing the three variables discussed above. The x-axis represented exposure (total number of vehicles entering the study section). It is calculated by multiplying the two-way ADT (in vehicles per day) by the project duration (in days). The RSAP analysis assumes a 50/50 split of the two-way ADT. The y-axis represents posted speed and influences the probability of encroachment and impact speeds (higher posted speeds result in higher encroachment and impact speeds). Figure 16 shows an example plot of benefit cost ratios for an outside lane and shoulder closure with minor encroachment into an adjacent open (to traffic) lane.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 68 It is apparent that in some cases benefit-cost ratios for equal values of exposure and speed are different. The encroachment rate versus ADT function is not a linear for certain ranges of ADT. Therefore, different combinations of ADT and project duration, which when multiplied together result in equal levels of exposure, will result in different encroachment frequencies and benefit cost ratios. In addition, the random component of the RSAP Monte Carlo algorithm results in differences in crash frequencies, severities and benefit-cost ratios. However, given RSAP’s convergence algorithm, differences resulting from this randomness are usually small. 35 40 45 50 55 60 65 70 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Exposure (106 vehicles) = Two-way ADT (veh/day) x Work Zone Duration (days) x 1/1,000,000 Sp ee d Li m it (m ph ) 0.12 0.12 0.17 0.21 0.36 0.36 0.40 1.06 1.41 1.31 1.86 2.64 3.24 3.95 0.84 0.94 1.26 1.44 2.14 2.83 2.95 0.83 1.04 1.36 1.58 2.37 2.96 3.50 0.53 0.71 0.66 0.93 1.32 1.62 1.98 0.24 0.22 0.79 0.72 0.73 0.65 0.71 0.49 0.41 0.33 0.33 0.25 0.23 0.15 0.37 0.28 1.19 0.98 1.09 0.94 1.07 0.71 0.62 0.48 0.50 0.42 0.35 0.31 0.44 0.41 1.50 1.44 1.31 1.38 0.99 1.11 0.67 0.79 0.50 0.51 0.30 0.36 0.82 0.98 3.01 3.62 2.76 3.63 2.21 2.67 1.58 1.70 1.03 1.53 0.72 1.15 1.23 1.41 4.51 5.51 4.14 4.05 3.32 3.28 2.37 2.32 1.54 1.98 1.08 1.37 2.16 1.97 2.75 1.96 1.88 2.02 1.48 1.43 1.64 1.00 0.96 1.16 0.75 0.84 0.99 0.45 0.62 0.69 0.66 0.56 0.71 0.11 1.250.820.960.620.18 0.21 0.27 0.32 0.35 0.34 0.69 0.84 1.25 1.03 0.53 0.54 0.63 0.61 0.47 0.37 0.60 0.48 0.47 1.0 1.16 0.80 1.66 1.78 Figure 16. Example plot of benefit-cost ratios for an outside lane and shoulder closure with minor encroachment into an open lane. The benefit-cost plots were divided into four regions representing the following ranges of B/C ratios: • (B/C ratio > 1.25); • (0.75 < B/C ratio ≤ 1.25); • (0.5 < B/C ratio ≤ 0.75); • (B/C ratio ≤ 0.5). These regions are illustrated in Figure 16. In cases where different B/C ratios were computed for equal levels of speed and exposure, the conservative estimate (i.e.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 69 higher ratio) was used. An attempt was made to have breaks lines separating the boundaries at round increments of speed (5 mph) and exposure (500,000 vehicles). The regions were then shaded and labeled for inclusion into Chapter 5 of the work zone design guidance (Appendix A). An example is illustrated in Figure 17. 35 40 45 50 55 60 65 70 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Exposure (106 vehicles) = Two-way ADT (veh/day) x Work Zone Duration (days) x 1/1,000,000 Sp ee d Li m it (m ph ) B/C ≤ 0.50 0.75 < B/C ≤ 1.25 B/C > 1.25 B/C = 1.0 Figure 17. Example of benefit-cost chart included in work zone roadside design guidance. The advantage of this approach is that it provides agencies and designers greater flexibility to establish policy or project level decisions. The disadvantage is that some agencies and individual designers may not find the guidance definitive enough. Consideration was given to labeling the regions with more definitive design-decision language. For example, the inclusion of notes such as “Barrier Study Optional” or “Barrier Not Normally Considered” were considered for the region with B/C ratios less than or equal to 0.5. However, such language would involve value judgments and might also mask the actual research results (which are estimated B/C ratios). Policy- and project-level decisions often (and properly) reflect numerous considerations, one of which may be cost-benefit analysis. The scenarios presented in Chapter 5, section 5.5, of Appendix A represent commonly occurring conditions for which a designer would have to choose whether or not to provide temporary concrete barrier. The scenarios were developed using MUTCD Part 6H Typical Applications, as well as input from NCHRP Project 3-69 panel members and other practitioners. Each scenario contains a short description, an illustration of the work zone layout and cross sections, and the design guidance resulting from the RSAP

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 70 runs. It should be noted here that in situations where there is a lane or shoulder closure with some encroachment on the remaining travel lanes, a design decision that must be made is how to distribute the remaining paved roadway for temporary lanes and shoulders. A number of combinations can often exist. For the tools used to develop barrier placement guidance, different combinations of lane and shoulder widths did not affect the resulting guidance in the referenced sections. Six scenarios and the resulting benefit-cost design aids are included in Appendix A: • Scenario 1, outside lane and shoulder closure for part-width construction on a four-lane divided highway; • Scenario 2, outside shoulder closure on a four-lane divided highway with minor encroachment; • Scenario 3, median shoulder closure on a four-lane divided highway with minor encroachment; • Scenario 4, bridge reconstruction with a temporary diversion/runaround on a two-lane, two way highway; • Scenario 5, separation of two-lane, two-way traffic on a normally divided facility; • Scenario 6, protection of a normally downstream barrier end for two-lane, two-way traffic on a normally divided facility. Scenarios 1 through 3 have several roadside condition similarities. For a specific ADT, duration and speed, one might expect: • For Scenario 1 to have the highest benefit-cost ratio for a barrier (compared to Scenarios 2 and 3) because all traffic is closer to the roadside hazards; • For Scenario 3 to have the lowest benefit-cost ratio for a barrier (compared to Scenarios 1 and 2) because 67 percent of traffic is located further away from roadside hazards than scenarios 1 and 2; • For Scenario 2 to have a benefit-cost ratio for a barrier somewhere between Scenarios 1 and 3 because 67 percent of traffic is located the same distance from roadside hazards as all Scenario 1 traffic and 33 percent of traffic is located further away. RSAP runs did not produce these results. For equal ADT and speed, benefit-cost ratios for Scenarios 2 and 3 were practically equal to each other and significantly greater than those for Scenario 1. Through subsequent testing and conversations with the RSAP development team, potential programming errors were detected. For purposes of this project, conservative estimates of benefit-cost ratios (resulting from the analysis of Scenario 2) were used to represent all three scenarios. Because it is a two-lane, undivided highway, Scenario 4 uses a different encroachment model than the other scenarios. Using this two-lane, undivided highway encroachment model leads to lower benefit-cost ratios at higher levels of exposure for many of the ADT and speed combinations that were run. This result is difficult to accept as valid and a modification of the encroachment model was made. The highest

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 71 encroachment rate for this facility type occurs for an ADT of approximately 5000 vehicles per day. Therefore, this ADT was used for all runs, and different levels of exposure were computed by only varying project durations. Scenario 5 posed a unique challenge. Attempts to model potential head-on collisions associated with this scenario were unsuccessful. Therefore, earlier work by Sicking and Ross (referenced in section 4.1.3) was used for the design aid related to this scenario. Several concluding points will be noted about the method used to develop the estimated benefit-cost ratios. RSAP is based on a probabilistic approach to roadside safety. Further, as outlined previously, there are numerous assumptions associated with RSAP’s development and its application to work zone scenarios. Consequently, the results it produces (including benefit-cost ratios) should be regarded as estimates of what would occur over many repetitions of the same conditions. Further, real world scenarios rarely conform to the exact conditions modeled. Therefore, the results shown should not be regarded as precise or always-accurate indication of the cost-effectiveness of a specific barrier placement. 4.1.6 Integration and Fusion to Develop Roadside Design Guidance The guidance in Chapter 5, Roadside Design and Barrier Placement, of Appendix A is divided into 6 sections. Section 5.1 is an introduction to roadside safety in construction work zones. It is based on Chapters 1 and 3 of the Roadside Design Guide and presents the underlying principles of roadside safety and design. These include the forgiving roadside concept and the prioritized treatment of hazards. It also contains some important distinctions between permanent roadway segments and construction work zones. Section 5.2 of Appendix A discusses the clear zone concept and its applicability to construction work zones. Several disadvantages and shortcomings of the clear zone concept are provided. These observations are based on statements from the Roadside Design Guide and from some state DOTs that provide work zone clear zone guidance. Nonetheless, the simplicity and practicality of the clear zone concept is recognized, and suggested dimensions are provided. The basis for these dimensions is guidance from Illinois DOT, which seemed to provide representative ranges of suggested clear zones (i.e., similar to guidance from other states, such as Indiana, Montana, and Oklahoma, that provided work zone clear zone dimensions). Section 5.3 contains a discussion on the identification of work zone roadside hazards that may require treatment or shielding. For the most part, designer experience and judgment are relied upon for identification of hazards. However, a list of hazards often present in work zones is provided, compiled from guidance of several states: Illinois, Indiana, Mississippi, Montana, and Oklahoma.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 72 Section 5.4 of Appendix A discusses roadside safety and economics and is largely based on Chapter 2 of the Roadside Design Guide and on NCHRP Report 492. The main discussion item is the use of benefit-cost analysis for roadside safety treatment decisions. Section 5.5 presents the results of benefit-cost analysis to develop barrier placement guidance for a series of construction work zone scenarios. Study methodology detail is provided in section 4.1.5, some of which is repeated in the Appendix A to provide sufficient background detail for a designer using the aids. Section 5.6 of the Appendix A presents a variety of other topics associated with traffic barriers and other roadside safety features in construction work zones. A number of the sections cross reference the Roadside Design Guide, primarily Chapter 9, which “describes the safety, functional, and structural aspects of traffic barriers; traffic control devices; and safety features used in work zones; and provides guidance on their application.” Information contained in the Roadside Design Guide is not repeated or summarized. 4.2 SPEED MODEL 4.2.1 Introduction to Artificial Neural Networks Artificial neural networks (ANNs) have successfully been employed by researchers over the past 25 years in solving a wide variety of engineering problems. However, it has only been recently that ANNs have found their way into the area of transportation safety and operations. ANN structure and methodology are loosely based on the biological nervous system, which consists of many interconnected neurons similar to the two displayed in Figure 18. Each neuron consists of a cell body, dendrites and axon. Signals are passed from the axon of one neuron to the dendrite of another through a connection point called the synapse. Memories are stored by changing the connection strength of the synapse. The cell body then sums and thresholds all incoming signals to produce a new signal that is sent out the axon (53). ANNs operate on a much smaller scale but use the same basic principles. As with the biological nervous system, ANNs consist of many interconnected but “artificial” neurons that weight, sum and threshold incoming signals to produce an output. Information is also stored within the strengths of the interconnections or weights. Figure 19 depicts a typical ANN architecture. The neurons, sometimes referred to as nodes, are usually arranged into what are known as layers. The network shown in Figure 19 has one input layer (not always referred to as an actual layer), a hidden layer, and an output layer. Neurons in a layer are typically connected to every neuron in adjacent layers through a connection weight. These weights determine the function of the network. Each node sums its weighted inputs and then applies an activation function, typically a sigmoidal activation function, to produce an output.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 73 Figure 18. Schematic drawing of biological neurons. Figure 19. General structure of a feed-forward ANN. Just as new memories are formed in biological neural systems through adjustments in the synaptic connection strengths, new memories are formed in ANNs by adjusting the weighted connections between neurons. This is typically done through some well established training procedures where the network is presented pairs of input/output data and an attempt is made to search for a global minimum on the error surface over the space of the network parameters or weight values. Figure 20 demonstrates the basic training process.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 74 Some of the advantages of using an ANN are: • No assumptions need to be made as to the form of the model; • It is capable of extracting non-linear variable interactions; • It is able to generalize from small training data sets. Figure 20. Network training (53). 4.2.2 Selection of Input Variables The first step in developing the speed profile model was identifying variables that may affect vehicle speed and can serve as candidate model inputs. Although a large number of predictor variables were considered, selection was based on model goals and measurement feasibility. Since the scope of this study was to determine how road geometry and work zone traffic control affects vehicle speeds for passenger vehicles and large trucks, vehicle interaction/car following variables were not considered. The variable list went through several iterations and involved inputs from both transportation and dynamic system researchers on the NCHRP Project 3-69 team. Both continuous and categorical variables were included. It included geometric and traffic control features as well as upstream speeds and distances between speed data collection locations. Variables not included would be controlled for as much as possible during the data collection process. The final set of model inputs are discussed in section 4.2.4. 4.2.3 Data Collection and Descriptive Statistics High-speed highways are defined as “roads and highways with free-flow operating speeds of 50 mph and higher.” Since this research involves construction work zones on high-speed highways, data was collected only for high-speed facilities.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 75 Furthermore, the scope of the speed model was limited to single lane closures (with traffic using the travel lane adjacent to the closed lane) and lane closures with median crossovers on four lane divided facilities. The following definitions apply. Median crossover: a construction work zone type used on expressways (including freeways) wherein: • The number of lanes in both directions are reduced; • At both ends, traffic in one direction is routed across the median to the opposite-direction roadway on a temporary roadway constructed for that purpose; • Bi-directional traffic is maintained on one roadway while the opposite direction roadway is closed. Single lane closure: a construction work zone type where one travel lane and any adjacent shoulders are closed to traffic. Data were collected in a total of 17 construction work zones; 11 single lane closures and 6 median crossovers in Pennsylvania and Texas. For crossovers, data were only collected in the travel direction containing the crossover. The work zone set-ups were “standard” lane closures and median crossovers. Anomalies that were designed to accommodate an uncommon situation were avoided. Data were collected during the times of day where lengthy queues did not form at any point throughout the work zone. The speed profile of a traveling vehicle is continuous in nature; therefore, an ideal model should use a continuous representation of this profile. Such an approach would require tracking the speed of many vehicles through the entire length of a work zone, with each vehicle having its own unique profile for the particular site. Available methods of data collection, however, make it difficult to capture this profile as a continuous function. Therefore, measured speeds were captured only at particular locations or “points” throughout a work zone site. At each work zone, 2 to 19 locations were selected for speed data collection. One location was upstream of the work zone, prior to the influence of any temporary traffic control. The remaining locations were located in the lane taper and the activity area. The lane taper was defined as the transition between the normal cross section and the work zone cross section where one lane and the adjacent shoulder were closed. Tapers were typically created by a series of channelizing devices such as vertical panels or drums. The work area comprised the remainder of the work zone from the lane taper to the termination point. Locations were selected to provide a range of conditions for roadway cross sections, roadside features and horizontal and vertical alignment. Locations where vehicle speeds appeared to be affected by the presence of entrance or exit ramps were avoided. At each location, approximately 200 free-flow speeds (defined as speeds of a vehicle with a headway greater than 4 seconds) were collected during dry, daylight conditions. In addition, traffic control plans combined with field observations were used to gather information on the following features:

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 76 • Travel lane width; • Right and left shoulder width; • Right and left shoulder type; • Presence of and offset to roadside objects (e.g. temporary or permanent barrier, work zone channelizing devices, other roadside conditions); • Radius of horizontal curve; • Vertical grade; • Rate of vertical curvature; • Posted speed limit; • Distance from the lane taper; • Cross slope. The final data set consists of 26,902 free-flow observations from 136 locations. The breakdown by state, work zone configuration, and location type (i.e. upstream, taper, activity area) is summarized in Table 9. Tables 10 through 15 summarize the descriptive statistics of the categorical and continuous variables collected at the upstream, lane taper and activity area locations.

F inal R eport for N C H R P R eport 581: D esign of C onstruction W ork Z ones on H igh-S peed H ighw ays C opyright N ational A cadem y of S ciences. A ll rights reserved. Table 9 Breakdown of speed data by location and work zone configuration for passenger cars (PC) and heavy vehicles (HV) Lane Closure Upstream Taper Work Area State No. of Locations No. of PC Observations No. of HV Observations No. of Locations No. of PC Observations No. of HV Observations No. of Locations No. of PC Observations No. of HV Observations PA 7 1096 328 6 824 399 23 2991 1593 TX 4 636 164 11 1598 577 11 1609 591 Total 11 1732 492 17 2422 976 34 4600 2184 Median Crossover Upstream Taper Work Area State No. of Locations No. of PC Observations No. of HV Observations No. of Locations No. of PC Observations No. of HV Observations No. of Locations No. of PC Observations No. of HV Observations PA 3 377 168 2 215 185 21 2290 1715 TX 3 398 202 4 475 325 41 4812 3334 Total 6 775 370 6 690 510 62 7102 5049 Total Upstream Taper Work Area State No. of Locations No. of PC Observations No. of HV Observations No. of Locations No. of PC Observations No. of HV Observations No. of Locations No. of PC Observations No. of HV Observations PA 10 1473 496 8 1039 584 44 5281 3308 TX 7 1034 366 15 2073 902 52 6421 3925 Total 17 2507 862 23 3112 1486 96 11702 7233

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 78 Table 10 Descriptive statistics of candidate categorical predictor variables Lane Taper (23 locations) Work Area (96 locations) Variable Categories Frequency Percent Frequency Percent Left 7 30.4% 9 9.4% Lane closed Right 16 69.6% 87 90.6% 50 4 17.4% 25 26.0% 55 2 8.7% 8 8.3% 60 4 17.4% 31 32.3% 65 7 30.4% 16 16.7% Posted speed 70 6 26.1% 16 16.7% no 6 26.1% 40 41.7% yes 2 8.7% 4 4.2% Police presence missing 15 65.2% 52 54.2% Permanent 23 100.0% 66 68.8% Roadway type Temporary 0 0.0% 30 31.3% Tangent 19 82.6% 52 54.2% Curve to the left 2 8.7% 27 28.1% Horizontal alignment Curve to the right 2 8.7% 17 17.7% Flat (-1 to 1) 12 52.2% 41 42.7% Upgrade 2 8.7% 21 21.9% Downgrade 1 4.3% 25 26.0% Crest curve 7 30.4% 4 4.2% Vertical alignment Sag curve 1 4.3% 5 5.2% Incoming grade 2 8.7% 0 0.0% Middle 1 4.3% 5 5.2% Outgoing grade 1 4.3% 2 2.1% N/A 15 65.2% 87 90.6% Location in vertical curve Missing 4 17.4% 2 2.1% None 15 65.2% 31 32.3% Drum 7 30.4% 7 7.3% Panel 0 0.0% 2 2.1% Guardrail 0 0.0% 4 4.2% Concrete barrier 1 4.3% 50 52.1% Traffic control device (TCD) to the left Opposing traffic 0 0.0% 2 2.1% None 7 30.4% 36 37.5% Drum 8 34.8% 17 17.7% Panel 1 4.3% 9 9.4% Guardrail 0 0.0% 9 9.4% Concrete barrier 4 17.4% 20 20.8% Traffic control device (TCD) to the right Other 3 13.0% 5 5.2%

F inal R eport for N C H R P R eport 581: D esign of C onstruction W ork Z ones on H igh-S peed H ighw ays C opyright N ational A cadem y of S ciences. A ll rights reserved. Table 11 Descriptive statistics of candidate continuous predictor variables Lane Taper (23 locations) Work Area (96 locations) Variable N Minimum Maximum Mean Std. Deviation N Minimum Maximum Mean Std. Deviation Length from taper (miles) 23 0 0.2 0.06 0.08 96 0.2 10.6 3.03 3.03 Radius of curve (ft) 4 2292 7640 4018 2448 44 1911 11480 5743 3198 Superelevation (%) 22 2 7.5 2.46 1.48 81 2.0 7.5 2.56 1.43 Incoming grade (%) 22 -3.22 3 0.39 1.35 96 -4.0 3.0 -0.33 1.77 Outgoing grade (%) 5 -3.5 -2 -2.96 0.61 7 -2.7 3.0 -0.18 2.41 Rate of vertical curvature (K) (ft/%) 7 247 615 364 127 9 150 500 258 128 Traveled way width (TWW) (ft) 23 12 24 17 4.6 96 11 16 12.44 1.29 Right shoulder width (RSW) (ft) 23 0 10 3 4.6 96 0 16 4.17 4.10 Left shoulder width (LSW) (ft) 23 0 8 3.7 3 96 0 36 3.23 4.40 Total paved width (TPW) (ft) 23 16 34 23 5.3 96 12 48 19.14 4.89 Left offset to TCD (ft) 8 0 5 1.1 1.8 65 0 48 3.91 9.44 Right offset to TCD (ft) 16 0 4 1.3 1.4 60 0 24 2.78 3.79

F inal R eport for N C H R P R eport 581: D esign of C onstruction W ork Z ones on H igh-S peed H ighw ays C opyright N ational A cadem y of S ciences. A ll rights reserved. Table 12 Descriptive statistics of 15th percentile speed (aggregated by location) for each vehicle type Aggregated 15th Percentile Speed for All Vehicles (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 49 68 63.53 4.30 Lane Taper 23 44 63 54.70 5.45 Work Area 96 29 60 51.69 4.95 Aggregated 15th Percentile Passenger Car Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 49 70 64.15 4.62 Lane Taper 23 45 66 55.66 5.94 Work Area 96 29 61 52.10 5.68 Aggregated 15th Percentile Truck Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 44 66 62.18 5.00 Lane Taper 23 44 61 53.18 5.04 Work Area 96 29 61 51.27 5.58

F inal R eport for N C H R P R eport 581: D esign of C onstruction W ork Z ones on H igh-S peed H ighw ays C opyright N ational A cadem y of S ciences. A ll rights reserved. Table 13 Descriptive statistics of mean speed (aggregated by location) for each vehicle type Aggregated Mean Speed for All Vehicles (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 54 73 68.29 4.16 Lane Taper 23 51 69 60.35 5.56 Work Area 96 37 66 56.42 5.30 Aggregated Mean Passenger Car Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 54 74 68.90 4.53 Lane Taper 23 51 71 61.36 5.91 Work Area 96 37 68 56.89 5.39 Aggregated Mean Truck Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 53 71 66.53 4.02 Lane Taper 23 49 66 58.21 5.11 Work Area 96 36 64 55.69 5.25

F inal R eport for N C H R P R eport 581: D esign of C onstruction W ork Z ones on H igh-S peed H ighw ays C opyright N ational A cadem y of S ciences. A ll rights reserved. Table 14 Descriptive statistics of 85th percentile speed (aggregated by location) for each vehicle type Aggregated 85th Percentile Speed for All Vehicles (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 59 78 73.00 4.20 Lane Taper 23 57 74 66.00 5.70 Work Area 96 44 72 61.11 5.11 Aggregated 85th Percentile Passenger Car Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 59 79 73.63 4.38 Lane Taper 23 57 76 66.81 5.85 Work Area 96 43 73 61.71 5.24 Aggregated 85th Percentile Truck Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 60 75 70.88 3.41 Lane Taper 23 55 70 63.16 5.06 Work Area 96 44 68 60.10 4.95

F inal R eport for N C H R P R eport 581: D esign of C onstruction W ork Z ones on H igh-S peed H ighw ays C opyright N ational A cadem y of S ciences. A ll rights reserved. Table 15 Descriptive statistics of standard deviation of speed (aggregated by location) for each vehicle type Aggregated Standard Deviation of Speed for All Vehicles (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 3.58 6.13 4.98 0.705 Lane Taper 23 4.17 6.60 5.49 0.552 Work Area 96 2.79 8.33 4.70 0.940 Aggregated Standard Deviation of Passenger Car Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 3.14 6.36 5.04 0.833 Lane Taper 23 4.45 6.68 5.46 0.648 Work Area 96 2.81 8.61 4.83 0.959 Aggregated Standard Deviation of Truck Speed (mph) Location Type N Minimum Maximum Mean Std. Deviation Upstream 17 3.31 6.48 4.18 0.731 Lane Taper 23 3.49 6.40 4.89 0.866 Work Area 96 2.18 7.82 4.27 0.972

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 84 After an initial analysis of the dataset, several geometric variables were eliminated from the model based on missing data and lack of variation. The following variables were eliminated due to missing data: superelevation (e), grade (G), and rate of vertical curvature (K). Several other variable categories (Other Soft Roadside Device Left and Right and Opposing Traffic for Roadside Device Right) were eliminated due to lack of observations. 4.2.4 Methodology for Development of ANN model The most common network architecture is a two-layer feed-forward network with sigmoid transfer functions in the hidden layer and linear transfer functions in the output layer as shown in Figure 21. The reason that this network is so often used is because it has been shown that it is capable of approximating any function to any degree of accuracy, depending on the number of hidden neurons (53). Figure 21. Two-layer feed-forward artificial neural network. The network output in this case is given by equation 4. 21 b)bpIW(LWOutput ++×Φ×= (4) Where 1 e1 2 x2 −+ =Φ − (5) A block diagram of the ANN model implemented in this study is given in Figure 22. A two-layer feed-forward network with two hidden neurons and one output neuron was used in this study. The output of the model is the speed of a vehicle, ν, as a function of distance, x, measured from the beginning of the work zone or lane taper. The model predicts speeds only for locations of “x” for which an input vector is defined. The ANN model predicts vehicle speed based on three inputs. The first input, u WZ (x), is a vector containing the geometric variables at the particular location. Some of these variables, like work zone type, are constant for a

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 85 particular site, while most variables change depending on the particular point within the site. The second input, ν(0), is upstream speed. This is the estimated speed of a vehicle prior to entering the work zone and is typically 2 to 3 miles upstream from the lane taper. The upstream speed is used in predicting all other speeds within the work zone. The final input is the previous predicted speed, which is fed back from the model output through a distance delay block. For the first speed predicted in a work zone, the previous predicted speed is the upstream speed, ν(0). It is important to note that distance to the previous predicted speed is included in u WZ (x). It is necessary to include this variable since data collection points were not equally spaced. Figure 22. Block diagram of the speed profile model. Before developing an ANN model, it is first necessary to transform the input vectors into a form that will be conducive to network learning. This was accomplished through a variable encoding process. A total of 31 network inputs were derived from the geometric database. The final list of model inputs is given in Table 16. Categorical variables were encoded using a binary representation which is typical in neural network implementation. For example, a variable containing N categories was represented using N separate binary inputs. Radius of horizontal curvature was inverted in order to represent the quantity within a finite range. Since the inverse of radius contained similar, but more detailed, information to the categorical variable horizontal alignment, it was not necessary to include both variables as inputs. For this reason, horizontal alignment was eliminated as an input. The Neural Network Toolbox in MATLAB (54) was used to develop the ANN model. Network inputs and targets were first normalized using the PRESTD command in MATLAB. The first step in training the network is separating the dataset into two groups: one for training the network and the other for testing the network. Because of the limited number of data points available, the testing dataset needed to be carefully selected such that it was a representative set. A total of five sample points was chosen for testing.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 86 Table 16 Final model inputs Predicted network outputs are compared to the actual measured speeds or target values. Network parameters or weight and bias values are adapted in the training process to minimize error. There are several different optimization algorithms that are capable of performing this operation. In this study, TRAINRP algorithm (resilient back-propagation) of MATLAB’s NN Toolbox was used. 4.2.5 Results The ANN models were developed for six different datasets (two each for cars, trucks and all vehicles) as listed below: • Mean of speed data for cars; • Variance of speed data for cars; • Mean of speed data for trucks; • Variance of speed data for trucks; • Mean of speed data for all vehicles combined; • Variance of speed data for all vehicles combined. Input variables (descriptions) choices and ranges Work Zone Configuration; discrete choices: Lane Closure or Median Crossover Upstream Speed (estimated or measured 85th percentile speeds upstream of work zone) any value between 48 and 72 mph Location; discrete choices: Taper or Within Work Zone Distance (location of analysis/prediction point measured from the taper) any value between 0 and 10.6 miles Posted Speed (posted speed at prediction point) discrete choices: 50 to 70 mph, inclusive at 5 mph increments Roadway Type; discrete choices: Permanent or Temporary R (Radius of curvature) any value between 1,191 and 11,400 ft; 99999 is entered to signify a tangent VA (Vertical alignment) discrete choices: Flat, Upgrade, Downgrade, Crest or Sag TWW (Traveled way width) any value between 11 and 24 ft RSW (Right shoulder width) any value between 0 and 16 ft LSW (Left shoulder width) any value between 0 and 36 ft TPW (Total paved width) any value between 12 and 48 ft RSDL (Roadside device on left) discrete choices: None, Drum, Vertical Panel, Guardrail, Barrier, or Opposing Traffic w/ No Separation Loffset (Left offset; the distance from the edge of the travel lane to the roadside device on the left side of the road) any value between 0 and 48 ft RSDR (Roadside device on right) discrete choices: None, Drum, Vertical Panel, Guardrail, or Barrier Roffset (Right offset; the distance from the edge of the travel lane to the roadside device on the right side of the road) any value between 0 and 24 ft

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 87 The results obtained using the mean speed datasets for cars, trucks and all vehicles are shown in Figures 23, 24, and 25. A summary of the mean square errors (MSE) obtained for all nine datasets is given in Table 17. 20 40 60 80 30 40 50 60 70 80 Measured Speed P re di ct ed S pe ed Best Linear Fit: A = (0.733) T + (15.3) R = 0.887 Data Points Best Linear Fit Perfect Prediction 0 50 100 150 30 40 50 60 70 80 Training Data Network Performance MSE= 7.0893 Training Point Number S pe ed (m ph ) Measured Speed Predicted Speed 45 50 55 60 45 50 55 60 65 Measured Speed P re di ct ed S pe ed Best Linear Fit: A = (0.927) T + (5.38) R = 0.96 Data Points Best Linear Fit Perfect Prediction 1 2 3 4 5 45 50 55 60 65 Testing Data Network Performance MSE= 3.7791 Testing Point Number S pe ed (m ph ) Measured Speed Predicted Speed Figure 23. ANN results for mean speed data for cars.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 88 30 40 50 60 70 30 40 50 60 70 Measured Speed P re di ct ed S pe ed Best Linear Fit: A = (0.734) T + (14.9) R = 0.885 Data Points Best Linear Fit Perfect Prediction 0 50 100 150 30 40 50 60 70 Training Data Network Performance MSE= 6.0341 Training Point Number S pe ed (m ph ) Measured Speed Predicted Speed 45 50 55 60 45 50 55 60 Measured Speed P re di ct ed S pe ed Best Linear Fit: A = (0.756) T + (14.1) R = 0.93 Data Points Best Linear Fit Perfect Prediction 1 2 3 4 5 45 50 55 60 Testing Data Network Performance MSE= 5.1169 Testing Point Number S pe ed (m ph ) Measured Speed Predicted Speed Figure 24. ANN results for mean speed data for trucks.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 89 30 40 50 60 70 30 40 50 60 70 Measured Speed P re di ct ed S pe ed Best Linear Fit: A = (0.717) T + (16) R = 0.869 Data Points Best Linear Fit Perfect Prediction 0 50 100 150 30 40 50 60 70 Training Data Network Performance MSE= 7.4771 Training Point Number S pe ed (m ph ) Measured Speed Predicted Speed 45 50 55 60 45 50 55 60 Measured Speed P re di ct ed S pe ed Best Linear Fit: A = (0.685) T + (17.3) R = 0.925 Data Points Best Linear Fit Perfect Prediction 1 2 3 4 5 45 50 55 60 Testing Data Network Performance MSE= 4.4445 Testing Point Number S pe ed (m ph ) Measured Speed Predicted Speed Figure 25. ANN results for mean speed data for all vehicles. Table 17 Mean square errors (MSE) with different datasets Model MSE - Train Set MSE - Test Set Cars – Mean speed 7.0893 3.7791 Cars – Mean speed variance 24.6589 1.6753 Trucks – Mean speed 6.0341 5.1169 Trucks – Mean speed variance 25.7865 12.3127 All – Mean speed 7.4771 4.4445 All – Mean speed variance 19.9121 4.8804 Cars – Mean speed 7.0893 3.7791 Cars – Mean speed variance 24.6589 1.6753 Trucks – Mean speed 6.0341 5.1169

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 90 4.2.6 Excel Implementation The spreadsheet then calculates the predicted speed profiles for 15th percentile speed, mean speed and 85th percentile speed and plots them on one graph. The 15th and 85th percentile speeds are calculated as follows: • 15th Percentile speed = Mean speed – 1.036 S.D; • 85th Percentile speed = Mean speed + 1.036 S.D. S.D. refers to the standard deviation that is calculated as the positive square root of the variance predicted using ANN. The underlying assumption here is that the speed distribution is normal and is symmetrical about the mean speed. There are three separate spreadsheets: one each for cars, trucks and all vehicles. A screenshot of the program can be seen in Figure 26. A more detailed snapshot is given in Figure 27. The Excel sheet has a protected area wherein the weight and bias matrices obtained from the MATLAB ANN model are input. The user is advised against any alteration of the values in that area, as it can affect the performance of the model. A user guide for the Excel model, along with the program CD, is included in Appendix B. Figure 26. Screenshot of speed profile model in Excel.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 91 Figure 27. Excel speed profile model detail. 4.2.7 Conclusions The goal of this part of the research was to develop a speed profile model that will enable designers to detect design inconsistencies in construction work zone designs for high-speed highways before their implementation. An ANN was selected for model generation because of the advantages it offers. Two advantages include the elimination of the need to guess the form of the model and the capability to automatically model most relationships. Input data for the model were collected from 17 work zones in Pennsylvania and Texas. A total of 119 sample points (excluding the upstream points) was obtained. Once developed, the ANN model was implemented in Microsoft Excel for ease of use. The model exhibits good prediction accuracies. However, accuracies can be improved further by collecting data points at closer intervals throughout the work zones and by increasing the sample size. Overall, this research has shown the potential that ANN models offer for future applications involving transportation safety. 4.3 PRELIMINARY AND DETAILED DESIGN OF SPECIFIC WORK ZONE TYPES AND FEATURES Guidance has been developed on many subjects. It is desirable that all guidance be based on a thorough understanding of how each decision will affect traffic safety and mobility and other outcomes of interest (e.g., speeds, driver comfort). Some design

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 92 factors have been the subjects of previous research, while many others have not. Therefore, guidance that provides for documented relationships among all design factors and performance is not achievable. As a practical matter, work zone design decisions are a routine part of transportation agency processes. The guidance used to make these decisions was regarded as baseline information that carries some level of empirical validation. However, guidance published by the various state DOTs is not uniform. The guidance on specific subjects sometimes varies, which may be attributable to setting (e.g., climate, terrain), service demands (traffic), organizational priorities or differing technical approaches. The scope of coverage also varies, presumably reflecting the needs and priorities of individual DOTs. Access to state DOT procedures was generally attained by survey instrument responses and procedures submitted with the survey or information available from an agency Web site. Additionally, several DOTs provided information as part of informal communications during the course of the research. The span and coverage of the guidance was identified early in the research. Priorities were identified using state DOT survey input and direction from NCHRP, leading to detailed studies on several subjects. For other topics, the general approach was to develop guidance on the basis of various sources (e.g., existing state policies and practices, published research and design principles). The purpose of this section is to document the origin and basis for areas of design guidance, designated by subheadings. 4.3.1 Work Zone Strategies and Planning It was found that work zone guidance publications of some DOTs provide information on the identification and evaluation of candidate work zone strategies, while others do not. Inclusion of this information was considered desirable and consistent with the scope of the research. Definitions and basic information for specific work zone configurations and mitigation strategies is provided. Definitions were developed by the research team for work zone types that appeared in state DOT design guides and literature. Terms are not always used in the same manner. For example, “detour” and “diversion” are sometimes used to describe the same work zone type. When different meanings are applied to the same term, the more common or clearly established usage was adopted. In the example cited, the MUTCD Part 6 distinguishes between detour and diversion. Several state DOTs (e.g., Connecticut, Illinois) use an identical matrix-formatted exhibit, “Identification of Feasible Work Zone Types,” presumably derived from Planning and Scheduling Work Zone Traffic Control (55). The work zone design guidance Exhibits 3-4 and 3-5 in Appendix A were developed as an extension of previous work, in the following manner. First, an effort was made to distinguish work zone types that reduce capacity from those that mitigate. For example, in practitioner discussions, a project work zone type might be referred to simply as a “detour” or alternatively a “diversion.” In these examples, it is clear that the permanent road is closed. However, in other cases a single term may not be self-explanatory. For example “lane constriction,” which is one of the work zone types listed in the Identification of Feasible Work Zone Types matrix (55), may or may not include capacity mitigation by maintaining the same

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 93 number of lanes. Another enhancement is the development of separate exhibits for two- lane and multi-lane facilities. This was provided to facilitate use by designers. Additionally, numerical rankings for the various strategies were included. These are subjectively established based on general impacts to traffic, cost and feasibility. The guidance document clearly states that the numerical rankings are not an indication of the appropriate choice for a specific set of conditions but rather are a general guide to establish an order or range of alternatives that should be evaluated using the factors provided in the same chapter. Several state DOTs (e.g., Illinois, Indiana) included information on one or more contracting strategies in work zone design guidance. It was concluded that this information may be useful to other agencies, even though some agencies may have policy or legal restriction against some strategies. The information included in the guidance was prepared after reviewing the FHWA Web site on innovative contracting (56) and state DOT guidance. 4.3.2 Controls and Principles Design controls are defined in the guidance as factors that lie outside the designer’s discretion but may affect the design process and the designed solution. The categories of controls and associated narrative description were developed by the research team. Highway design in general and work zone design in particular are not exact sciences, governed entirely by deterministic processes. However, a body of knowledge has evolved, including principles that apply to many specific design areas. These principles are largely empirical and rise above complex subsystems through an often simple expression. An example is the forgiving roadside concept, which implicitly embraces the cumulative performance of driver, infrastructure and vehicle subsystems. Yet the guiding principle has continued relevance and benefit in the design process. The Green Book includes an explanation of (driver) expectancy. The guidance provides a principle of design consistency that is based on expectancy. Although these topics are closely related, design consistency is generally recognized as an objective of the design process, while expectancy describes a human factor. The principle of primacy is described and applied in the guidance in a manner similar to that in the Green Book. Guidance on speed management and consideration of speed in design decisions was developed by the research team based on several sources. General information on speed management was obtained from the FHWA Web site (57). The term and definition for “work zone design speed” was developed after reviewing how various DOTs consider speed in work zones, the role of “design speed” in the Green Book, and the MUTCD Part 6 provision on work zone speed limits. Speed, and especially speed in work zones, has been a controversial subject for some time. The research team attempted to develop guidance that is consistent with general speed management doctrine, reinforces the MUTCD, is generally consistent with the practices of some DOTs, and is useful to practitioners at the project level.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 94 Linkages were identified between construction work zones, desirable speed behavior, work zone design, and implementing speed management and control techniques. This guidance (section 2.2.4 of Appendix A) extends beyond any existing agency guidance reviewed during the research. It was developed as a step toward reconciling the inconsistencies among design, regulatory and operating speeds that pervade work zones and permanent facilities. Guidance on roadside safety, roadside design and barrier placement in construction work zones builds on previous research and design guidance, notably that reflected in the Roadside Design Guide and state DOT design practices, as outlined in section 4.2 of Appendix A. An extensive effort was made to the development of objective guidance for commonly occurring work zone design scenarios. 4.3.2.1 Sight Distance Development of guidance related to sight distance presented a quandary. There is no doubt that some amount (length) of sight distance is needed to avoid collisions. Sight distance criteria based on speed, as in the existing Green Book approach for stopping sight distance, seems appropriate. Several studies investigating a possible relationship between available sight distance and crash rates have been conducted and are summarized by Fambro et al. (58) in their research report that is the source of the stopping sight distance criteria in the current Green Book. That report concluded that “in the sight distance ranges studied, limited stopping sight distances had no discernable effect on accident frequency or rate.” However, one of the studies (59) reported that crash frequencies on crest vertical curves with sight distances of less than 300 feet were more than 50 percent higher than on crest vertical curves with very long sight distances. No research on the relationship between sight distance in work zones and safety was identified. Practice among state DOTs in this area is divided. Half (16 of 32) of the state DOT respondents indicated that a stopping sight distance criterion is used in work zone design. Under these conditions, specific stopping sight criteria were not recommended. Instead, the design guidance includes a discussion of sight distance and the criteria used by some agencies. The literature on this subject leads to the conclusion that extended sight distance approaching and within work zones is desirable from an operations perspective. Safety considerations also point to some minimum sight distance need, but not necessarily as a function of a speed parameter. Many state DOTs use stopping sight distance criteria based on the Green Book and corresponding to work zone design speed. These values are not unreasonable for use in designing work zones but do not necessarily represent the minimums that can be accepted. Based on these considerations the design guidance recommends at least 300 feet of sight distance for construction work zones with work zone design speeds of 40 mph and greater. For work zone design speeds of less than 40

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 95 mph, Green Book stopping sight distance values are recommended. Green Book values for driver eye height and object height are recommended. 4.3.2.2 Roadway Surface and Cross Section Exhibit 2-2 in the design guidance document is an adaptation of Montana DOT’s guidance. As outlined in section 3.2.4.2 of this report, several states (Connecticut, Montana, North Carolina, Texas, Wisconsin) selectively use lower-level road surfaces for low levels of exposure. The Montana DOT guidance accounts for both traffic volume and duration and is therefore considered the most appropriate format. The values also appear appropriate in the context of practice by other states. Roadway and shoulder cross slope guidance is based on permanent road design. 4.3.2.3 Horizontal Alignment-Superelevation Horizontal alignment design in work zones is generally limited to temporary roadways, such as diversions and median crossovers. As outlined in section 3.2.3 of this report, a number of alignment design practices and corresponding superelevation distribution methods are in use by various states. Green Book superelevation distribution Methods 2 and 5 are the most common. Both of these methods are identified in the guidance document as appropriate for designing curve-superelevation relationships in construction work zones on high-speed highways. Minimum radii values for work zone design speed and superelevation rates were computed using Method 2 and are provided as design aids. An example was developed demonstrating the use of the Method 2 design aid and its application to negative superelevation (adverse cross slope). 4.3.2.4 Vertical Alignment Guidance on maximum grades was developed primarily from a review of state DOT work zone design guidance, which is often based on design of permanent roads. Vertical curve designs for permanent roads are based on stopping sight distance criteria. Since specific sight distance criteria are not recommended by the guidance, it follows to not recommend minimum vertical curve lengths on that basis. However, the guidance references the influence of limited sight distance on operating speed and suggests that designers consider this effect in designing crest vertical curves. The guidance contains information from the Green Book on how to design a sag vertical curve for comfort, which is an appropriate basis for design absent another controlling consideration (i.e., sight distance, drainage). 4.3.3 Detailed Guidance by Work Zone Type Each work zone type is unique and poses distinctive design challenges. This uniqueness prevents any set of general principles from being complete or fully applicable to all design features for all work zone types. Therefore, specific design guidance is organized by work zone type. Much of the guidance relies on general principles, and

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 96 these provisions are referenced, rather than repeated or summarized, when applicable. The following discussion outlines the basis for work zone types and strategies. 4.3.3.1 Diversion Design of a diversion, a temporary roadway, involves the typical roadway design decisions. Guidance on horizontal alignment and superelevation, vertical alignment, and roadway surface for diversions references general guidance provisions that are considered applicable. Guidance on minimum roadway width was developed in consideration of state DOT guidance and general design-related safety principles. Permanent rural road design criteria, including those for minimum traveled way and shoulder widths and minimum widths for new and reconstructed bridges and existing bridges to remain in place, were also reviewed as bases of reference. The DOTs for Illinois, Indiana, Mississippi and North Carolina provide guidance for traveled way and shoulder widths for diversions. The range of recommended traveled way widths for conditions that would exist on a two-lane diversion of a high-speed facility vary from 18 to 24 feet. The guidance of two agencies does not include consideration of traffic characteristics (volume, mix) or curvature. One agency considers volume, and one considers both traffic mix and curvature. (For curve radii below 400 feet, recommended traveled way widths extend above 24 feet) The design recommendations for diversion shoulder widths by the same agencies vary from 2 to 8 feet. One agency recommends 6-foot shoulder widths for all diversions. The other three agencies recommend ranges of values: in two cases the specific value is based on traffic volumes; in the other case, the 2- to 4-foot range is simply stated. Diversions are temporary facilities that typically exist for a period of several months rather than several days. Traveled way width and exposure have been related to safety on permanent two-lane rural highways. For these reasons, recommended minimum widths are based on traffic volumes. For a two-lane bi-directional roadway, a minimum 22-foot traveled way width is recommended for lower volumes. Higher volumes warrant wider cross sections, especially shoulders that provide for recovery and disabled vehicle refuge. The values in Exhibit 4-5 of the design guidance are recommended values for traveled way and roadway (traveled way plus shoulder) widths. They are not regarded as minimum values. Feedback on the draft recommendations indicates that values below those listed as frequently used, and without reported adverse safety experience. Exceeding minimum values is not discouraged. 4.3.3.2 Lane Constriction Guidance for lane constrictions was developed after reviewing responses to the survey of state DOTs and considering factors that affect safety and operations. An additional objective was to establish a framework that provides guidance based on combinations of factors that are often identified individually. A research study (10)

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 97 summarized in Chapter 2, section 2.2.4, reports on crash rates for projects with and without reduced lane widths during construction. The results suggest reducing lane widths increase crash rates. A relationship between crash rate and magnitude of the reduction or the reduced lane width was not reported. Lane constrictions reduce the width of the traveled way and are therefore inherently undesirable from a safety and operations perspective. They are also an appropriate work zone type selection under certain circumstances. As outlined in section 3.2.4.1 of this report, nearly every state uses reduced traveled way widths under certain conditions. Factors that are often considered include: • Traffic volume and composition (high volumes and high percentages of heavy vehicles, and truck network route designation, weigh in favor of wider minimum width); • Facility type (higher minimum traveled way widths often pertain more to divided highways than to undivided facilities); • Constraint (the effect of a constraining feature [e.g., temporary barrier curb, structural element] along one or both borders of a traveled way influences driver performance). Several other factors were mentioned less frequently, including design speed, curvature, and the length (distance) and duration (time) of the lane constriction. One state (Colorado) also uses curvature and design speed, in addition to truck traffic. Most state DOT survey responses and guidance documents identify factors but do not indicate how the factors are considered in combination. All of the factors listed above are considered appropriate in design decisions and hence as bases in design guidance. Because of the many different combinations and uniqueness of each application this is understandable. However, a framework that provides guidance on the basis of several pertinent factors is desirable. Exhibit 4-6 of Appendix A was developed as an example of how various factors can be considered in combination. The tabulated values and accompanying notes are not based on safety or operational analysis. The exhibit is intended to illustrate how an agency might choose to establish guideline values. Guidance on the placement of temporary barriers within lane closures with minor encroachments was developed, as outlined in this report in sections 5.5.1, 5.5.2, and 5.5.3 of Appendix A. 4.3.3.3 Median Crossover Median crossovers are a common freeway work zone configuration comprised of temporary and permanent roadway elements, with design decisions required for each. General guidance provisions for work zone design speed selection, sight distance, vertical alignment, horizontal alignment and superelevation are considered applicable and referenced.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 98 The design guidance of Illinois, Indiana and Wisconsin DOTs all provide for a 16-foot traveled way width. Oklahoma DOT’s guidance calls for a 12-foot minimum traveled way on median crossovers. A width at or approaching 16 feet is considered appropriate because of the curvature, speeds and traffic customarily associated with freeways. No research exists to indicate the safety and operational consequences of variations in temporary work zone roadway and travel lane widths. The Illinois, Indiana, North Carolina and Wisconsin DOTs provide the same shoulder on both sides of a crossover; shoulder widths in conjunction with travel lane vary from 2 to 5 feet. Sample project plans were reviewed with shoulders up to 7 feet in width. The guidance recommends consideration of a wider right shoulder to reinforce its customary role as the location of refuge and to reduce potential conflict with temporary barriers. The guidance defers to agency experience. The design recommendations for multi-lane temporary roads, including shoulders, are based on limited published guidance. No research has been published specifically on safety consequences of multi-lane traveled way or shoulder widths for temporary roads in work zones. Guidance on the placement of temporary barriers within median crossovers was developed, as outlined in sections 5.5.5 and 5.5.6. 4.3.3.4 Use of Shoulder Connecticut, Illinois and Indiana DOTs provide guidance on this work zone type and were used to identify key issues. Some of the specific agency practices (e.g., replacement of shoulder pavement) were outlined in the guidance as possible action rather than a requirement. The research team expanded the scope of the guidance to address consideration of existing rumble strips and roadside design. The use of selective exclusion signs assigning heavy truck traffic to permanent traveled way lanes (i.e., prohibited from using shoulder as temporary lane) was gleaned from Illinois DOT guidance and was outlined in the guidance as an alternative, but one requiring enabling legal authority. The effects of relocating traffic closer to roadside hazards common to permanent roadways (e.g., culvert ends) was modeled with RSAP. Although more annual crashes with the hazard could result when a shoulder is converted to a travel lane, the short exposure time of one year (and often less for work zone durations) resulted in a benefit- cost ratio smaller than in the 25-year permanent roadway analysis (i.e., less than 0.5). These results indicate that it is not cost effective to install temporary guardrail to shield a hazard that does not justify shielding under permanent roadway conditions, even if traffic is shifted to the outside shoulder.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 99 4.3.3.5 Interchange Ramps One study on safety at interchanges within work zones was identified and is summarized in section 2.2.3 of this report. The conclusions are not useful in developing interchange ramp design work zone design guidance. The results of the state DOT survey documented in section 3.2.6.4 of this report show a wide variation in practice related to entrance ramp acceleration lane length within construction work zones. The range is illustrated by the state DOTs of Arkansas, Michigan, Oregon, New Jersey and Connecticut. As a result of this wide variation in practice and the absence of any established relationships between design of work zone acceleration lane lengths and safety, the formulation of guidance is left to operational considerations and judgment. The guidance describes the practices of Michigan and Oregon, without attribution, which are characterized as “rules of thumb.” The MUTCD Part 6 provides examples of temporary traffic control (TTC) at entrance ramps within work zones. Maryland State Highway Administration has developed guidance on TTC (i.e., stop and yield control with related signage) for acceleration lane length. A summary of the Maryland information is included in the work zone design guidance document produced in this research (Appendix A). Guidance pertaining to exit ramps is based on permanent road guidance. Dimensional values for taper and parallel deceleration lane lengths were obtained from New Jersey and Wisconsin DOT survey responses. 4.3.3.6 At-Grade Intersections The guidance was developed primarily on the basis of avoiding and managing traffic conflicts, using general strategies such as relocation, channelization and TTC. Several specific strategies (e.g., changing control type from yield to stop) were obtained from state DOT guidance documents. 4.3.4 Ancillary Design Features The work zone design guidance of several state DOTs addresses drainage in a general manner to determine that the same general practices (e.g., rationale formula, design charts) used for permanent roads are also used for general practice. The Florida DOT Drainage Handbook, Temporary Drainage Design (60) and AASHTO Model Drainage Manual (61) were the primary references for the guidance. The example of using a design storm with 10-year recurrence interval as the basis for designing a temporary bridge is based on Missouri DOT’s design guidance. The guidance on slotted drains was developed from field observations and review of the Texas DOT Web- accessible Hydraulic Design Manual (62). Design guidance for Enforcement Pullout Areas was developed by the research team based on other ongoing research on the same topic, separate from this project.

Final Report for NCHRP Report 581: Design of Construction Work Zones on High-Speed Highways Copyright National Academy of Sciences. All rights reserved. 100 A characterization of use and spacing of emergency turnouts was based on responses from 12 state DOTs to the survey. The schematic geometry layout is based on the guidance of New York and Wisconsin DOTs and a Pennsylvania DOT example project. Information related to the use of screens was developed from responses to the survey of state DOTs. Guidance on lighting was developed based on the policies of Indiana and Illinois DOTs and the cited research and reference publications. Information on rumble strips that supplements the MUTCD was based on research publications (63,64) and field observations.

Next: Chapter 5 Conclusions and Recommendations »
Design of Construction Work Zones on High-Speed Highways Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 581: Design of Construction Work Zones on High-Speed Highways explores an approach for the selection of an appropriate construction work zone type; offers suggested guidance for the design of geometric features, including horizontal and vertical alignment, cross-sectional features, and barrier placement; and examines a variety of ancillary features such as drainage systems, lighting, and surface type. The contractor’s final report on the research activities used to develop NCHRP Report 581 has been published as NCHRP Web-Only Document 105. As part of the research associated with this activity, a work zone prediction model and user's guide was created to help estimate free-flow vehicle speeds through two types of construction work zones on four lane freeways--single lane closures and median crossovers.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!