Estimating Life Expectancies of Highway Assets, Volume 1: Guidebook (2012)

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

11 When implementing any kind of decision support tool, good planning goes a long way to ensure that the tool will produce information that is reliable, useful, and relevant. It is easy, and all too common, to develop models that have considerable engineering and economic merit, but whose outputs never affect the management and political decisions that determine how money is allocated. It is therefore useful to list the benefits of transportation asset management. Reasons why senior management would be interested in the products of analytical tools such as life expectancy analysis include the following: • Credible long-term view. If procedures are in place to ensure that the inputs and analysis are routinely tested, adjusted, and validated to agree with real life, the life expectancy analysis can provide a useful and politically neutral way of comparing alternative policies and programs having long-term impacts. • Basis for transparency and accountability. Credible performance measures help all stakehold- ers to verify that promised project benefits are actually realized. • Means to specify the desired level of service. While the general relationship between funding and performance is widely appreciated but vaguely understood, the use of quantitative per- formance measures makes it possible to specify precisely how much performance is wanted and can be afforded. • A way to isolate the effects of traffic/demand growth and deterioration. Analysis tools such as life expectancy analysis help agencies and stakeholders understand the long-term investments necessary to maintain a desired level of service in the face of traffic growth and deterioration. • Maximize the benefits of infrastructure preservation. The ability to proactively estimate the effects of investments assists managers in balancing resource allocation to maximize network- wide performance delivered to all transportation system users. • Improve agency competitiveness for funding. Credible analytical tools give senior managers a competitive technique to use in funding negotiations. • Build constructive political relationships. Performance measures such as life expectancy provide a common language for communication and provide a basis for managers and outside stake- holders to work as a team to address deterioration and traffic growth, to best serve the needs of their shared customers. The planning process described in this chapter is condensed from the much more detailed presentation in the AASHTO Transportation Asset Management Guide (Gordon et al. 2010). By understanding the motivations of senior managers and stakeholders, the implementer of life expectancy analysis tools is in a better position to select and design business processes and analysis methods that will ensure that the results are credible and useful. C h a p t e r 2 Plan for Implementation: How to Plan Life Expectancy Models

12 estimating Life expectancies of highway assets 2.1 Documenting Business Processes Often the demand for analysis techniques originates with a single person or organizational unit that needs the information, but fulfillment of that demand necessitates the cooperation of many others in the organization who (1) might not understand whether or not they would ben- efit and (2) are already engaged in important duties. One of the most important implications of this insight is that all of the business units that might use the information are potential beneficia- ries and potential allies in advocating for the use of a new analytical tool and the accompanying change and improvement. A productive way to improve implementation success is to systematically identify the poten- tial partners, using a business process analysis (Jacobson 1995). Figure 2-1 shows an example. The idea is to show all the activities that the agency undertakes that either may benefit from life expectancy information or that affect the quality of the information and then to connect the boxes with data flows that are potentially relevant. This need not be a formal undertaking, but it does have strategic importance by indicating the people who could help or hinder implementa- tion success and then guiding the preparation of the list of desired reports and tools. In a corporate environment, it is not unusual for an analytically inclined engineer to team with a people-oriented product manager to secure the necessary support and resources. A diagram like Figure 2-1, a map of the contacts that need to be made, might be drawn on a napkin or written into a memorandum. By following the connecting lines, it is easy to trace the flows of data and see how better information can be available to each player and affect his or her decision-making. 2.2 Planning the Change Strategy Implementation of asset management tools, such as life expectancy estimation techniques, is a process of change. In an organization, change can be viewed with apprehension or with opposi- tion. It is important to recognize that change can have both positive and negative effects on each Figure 2-1. Example business process analysis.

plan for Implementation: how to plan Life expectancy Models 13 employee. Change management is often a process of engineering the effects of the change so that, from each person’s perspective, the positive outweighs the negative. Improved asset man- agement information often implies increased accountability, which can be especially alarming. Change leaders have to be especially sensitive to these fears and actively try to mitigate them. Successful organizational change to accommodate the new life expectancy estimation tool would require at a minimum the following activities (Gordon et al. 2010): • Convince employees of the need for and benefits of the change. For senior managers, the list at the beginning of this chapter is a helpful starting point. • Create a change leadership coalition, consisting of people who may benefit from the change and can articulate the benefit to the agency or the customer. Share the leadership duties and encourage creative input, even constructive disruption. • Develop a vision of the end result after the changes and the strategy needed to get there. • Communicate the vision regularly. • Take actions consistent with the vision. • Ensure that people are involved and are empowered to make changes consistent with the vision. • Reinforce the change effort with short-term successes. • Keep the focus on the change effort. • Anchor new approaches into the culture. Successful change is incremental and measured. If implementation of the life expectancy estimation tool is a part of process changes in the organization, as often will be the case, the user of this guide should follow the steps presented in Chapter 1 to determine where to start and how far to go with the implementation process. Each increment of change depends on the successful completion of the previous round of changes, with sufficient time allowed for the new capabilities and thought processes to sink into the culture. An agency in a relatively immature state of asset management may require several years to implement all the techniques described in this guide. 2.3 Listing Desired Reports and Tools The logical sequence of events in planning a life expectancy modeling capability change fol- lows a natural pattern, from general goals to a specific work plan (Figure 2-2). It is important to follow such a plan, rather than jumping directly to writing a spreadsheet or computer code. Such a plan ensures that the product will be relevant to as many people as possible in the agency and that the product will be valuable and used. Once the potential users and business processes are identified and the desired applications documented, it becomes possible to make a more specific list of the tools needed. At first, it is very likely that the most relevant tools will be spreadsheet models, which feed off of one or more asset management databases. For agencies lacking inventory and condition databases for certain types of assets, the first tools will likely involve databases. Figure 2-2. Sequence from general goals to specific work plan.

14 estimating Life expectancies of highway assets 2.3.1 Data Storage For most agencies at various stages of maturity and for most types of assets, simpler databases and applications are best. For data storage, consider using a desktop database (e.g., Microsoft Access) or a small network database (e.g., Microsoft SQL Server Express). For agencies hav- ing a mature data management infrastructure, consider working within that infrastructure to take advantage of the technical support. If the agency has a pavement, bridge, or maintenance management system in place that is working well, consider adding onto that database, rather than starting a new one. Asset management databases of the kind needed for life expectancy analysis are not large or complex, and many parts may already exist in the agency. For even the most sophisticated appli- cations described in this guide, the basic databases are as follows: • Geo-referencing database (usually the agency GIS) • Traffic count database (often included in the GIS) • Crash database (often maintained outside the transportation agency) • Asset inventory • Asset condition (may be a time series of inspections or surveys for each asset) • Asset vulnerability to natural and man-made hazards (may be a time series) • Climate condition database (often maintained outside the transportation agency) • Soil characteristics database (often maintained outside the transportation agency) For different types of assets, the inventory, condition, and vulnerability assessment databases may be located or maintained at different divisions or units in the agency. For example, there may be separate databases for pavements, bridges, other structures (such as tunnels, culverts, sign structures, signal mast arms, high-mast light poles, and retaining walls), signs, traffic sig- nals, pavement markings, guiderails, curbs and sidewalks, and buildings. Other databases such as those storing climate condition and soil characteristic data can be accessed through federal agency websites. The NOAA maintains various climate and extreme event data, most of which can be accessed and downloaded at no cost (http://www.ncdc.noaa.gov/oa/mpp/freedata.html). In this report, climate data such as average annual temperature, precipitation, and freeze-thaw cycles were found to be significant for predicting the service lives of culverts and bridges. Average wind speed was also found to be a significant factor in predicting traffic signal life. Most of the data can be downloaded at the climate division level, which groups geographically neighboring counties with similar climates (e.g., Figure 2-3). Similarly, soil data are maintained by the Natural Resources Conservation Service (NRCS). Soil attributes, such as corrosiveness and frost action potential (ranked from no potential to high potential), significantly affect the asset life of culverts and bridges because of the deterioration of the below-ground components of these structures. The NRCS database contains data on rel- evant soil attributes and other properties for soils located within each soil survey area (generally the size of a county) by depth (http://soildatamart.nrcs.usda.gov/). To analyze the significance of climate, soil, and other geographic properties, GIS applications are particularly useful for data storage, with each property having its own layer. For exam- ple, Chase et al. (1999) discusses how to add GIS spatial data to National Bridge Inventory (NBI) data. Certain types of assets could be managed as groups, rather than as individual facilities. For exam- ple, all the pavement markings on a segment of road, or even a corridor, could be inventoried and managed as a single unit. This approach works best if all the markings in the group have the same age, same material, same traffic volume, etc., so that they will have uniform life expectancy.

plan for Implementation: how to plan Life expectancy Models 15 2.3.2 Foundation Analysis Tools Once the basic data storage tools are established, consider selection of the analysis tools next. For life expectancy analysis, it will often be sufficient to have two sets of tools that make up the foundation of life expectancy analysis: • A network-level model that computes typical life spans for entire classes of assets using gen- eralized parameters; and • An asset-level model that computes life expectancy for each asset individually using its age, condition, and other characteristics, often using the network-level model as an input. Both of these types of tools are addressed in Chapter 4, with example applications in Chap- ter 5. None of the methods described in this guide are outside the capabilities of a spreadsheet model so these tools should not be considered major software investments (Figure 2-4). The methods and examples described in later chapters frequently refer to spreadsheet functions for statistical calculations. Using spreadsheet software is often the easiest way to implement these models and results in models that are fast, reliable, and inexpensive to develop and maintain. 2.3.3 Applications and Reports An efficient way to determine the desired applications and reports is to interview each of the potential users of the information identified in the earlier planning steps. The list of potential Figure 2-3. United States Climate Divisions (http://www.cpc.noaa.gov/).

16 estimating Life expectancies of highway assets Inspection pairs Condition - start of year Condition - end of year Improvement in condition Road Insp Condition state Condition state Condition state segment Year 1 2 3 4 1 2 3 4 1 2 3 4 RS0028 2004 92 8 0 0 82 17 1 0 -10 -1 0 0 RS0028 2005 82 17 1 0 68 27 4 1 -14 -4 -1 0 RS0028 2006 68 27 4 1 58 32 9 1 -10 -5 0 0 RS0028 2007 58 32 9 1 48 37 11 4 -10 -5 -3 0 RS0028 2008 48 37 11 4 46 35 12 7 -2 -4 -3 0 RS0028 2009 46 35 12 7 37 39 14 10 -9 -5 -3 0 RS0028 2010 37 39 14 10 32 37 19 12 -5 -7 -2 0 RS0061 2005 100 0 0 0 84 16 0 0 -16 0 0 0 RS0061 2006 84 16 0 0 78 19 3 0 -6 -3 0 0 RS0061 2007 78 19 3 0 67 27 5 1 -11 -3 -1 0 Figure 2-5. Example of a spreadsheet-based report. applications in Section 1.3, or a similar list tailored to the specific agency, will help to stimulate discussion. Most of the potential users of the information will prefer to receive periodic reports, on paper or as PDF files. This is the simplest approach for them so they need not remember how to use a spreadsheet or other software tool. Others will require a spreadsheet file (Figure 2-5), a system of related spreadsheet files, or a user interface because they may need to sort or filter the data they are working with or may need to enter or modify data as part of their decision- making responsibility. Figure 2-4. Example of organizing the foundation analysis.

plan for Implementation: how to plan Life expectancy Models 17 Some of the design variables to consider when determining the desired reports are as follows: • Filtering—Some users will want statewide reports, while others will need to see only a subset of the asset inventory (e.g., for a particular district, ownership, or asset type) to match their responsibilities. Certain reports may need to be filtered according to the year or time frame when assets are forecast to reach end-of-life or some other milestone. • Aggregation—Certain reports should list assets individually, while others will list only groups of assets. In life expectancy analysis, it is especially useful to group assets into cohorts that are geo- graphically close (due to similarities in climate and soils) and reach their end-of-life at about the same time. The life expectancy analysis is very helpful here in grouping facilities into projects. • Subject matter—Try to tailor reports to each user or user group to fit the exact subject matter they are concerned with. Do not crowd too much information into one report. Ask the users what is relevant, rather than including everything that seems like it might be relevant. • Sorting—Make sure the order of presentation of items in the report is logical for the end- user’s purposes. The best way to ensure the best presentation is to ask the intended users. Often it is useful to sort items in a report according to their urgency for the end-user’s atten- tion. Sometimes it is even necessary to make up a priority criterion that is intended for just one user or user group to provide a value on which to sort. For example, if a user wants to emphasize assets that expire soon as well as assets that have particularly high vulnerability to hazards, then it may be necessary to create a criterion that is a combination of these two (or more) data values. Other sorting criteria that are commonly used include asset identifiers, geographic location, current condition, or performance indices. • Graphics—Most end-users find graphs helpful, and life expectancy analysis provides good opportunities for useful and creative graphics (Figure 2-6). Chapter 5 has examples of relevant types of graphs, all of which can be produced by common spreadsheet and report-writing tools. 2.4 Defining the Work Plan and Resource Needs Particularly when working with commonly available software, constructing reports is relatively fast and inexpensive, so do not hesitate to plan for a large number of them. During the develop- ment phase, plan to interview users for detailed specifications, then produce a prototype report 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Pr ob ab ili ty Age Cumulative This year Average Figure 2-6. Example graphical output, showing uncertainty in life expectancy.

18 estimating Life expectancies of highway assets right away (e.g., within a week), ask for feedback, and then modify the report, again within a week. It is important to keep the end-user’s attention focused on the report until it is completed and ready to use. Chapter 1 lays out the main work plan tasks in developing life expectancy applications. An example work plan might be as follows: • Task 1. Define the scope of the analysis and the needs to be served within the time frame of the project. • Task 2. Develop an implementation plan. • Task 3. Define the performance metrics and analysis concepts. Determine data requirements and ascertain how the necessary data are to be obtained. In some cases this may necessitate the launching of new data collection processes, especially for assets other than pavements and bridges, where many agencies have minimal data. Some database development or modi- fication may be needed in this task. Create mockups of tools and reports to be developed in subsequent tasks. • Task 4. Develop the foundation tools for computing life expectancy for all the asset types within the scope of the project. In many cases, some research or statistical model estimation work may be needed within the scope. Plan to develop a working prototype of each analysis, solicit feedback from users, and then refine the prototype. Document the results in the form of a “User’s Manual” or “Technical Memorandum.” • Task 5. Build applications that put the new models to work in real business processes. In some cases, the development work may entail modifications to existing systems, especially pavement, bridge, and maintenance management systems. When a new application is needed, consider using media that facilitate prototyping and rapid development. It is often much easier to attach a separate spreadsheet model or report to a management system database than to try to modify the management system itself. • Task 6. Ensure that the products of the work have sufficient long-term support. Monitor and evaluate the use of the products and plan for further refinements. Be confident of the results and communicate this confidence to stakeholders. One of the basic rules of successful change management is to achieve early successes as a means of building and maintaining support. If certain asset types may require an extended work plan duration, perhaps because new data collection is required, then plan to develop other asset types in parallel that have readily available data. Plan a sequence of regularly spaced rollouts to keep interest high while buying time to complete the more difficult parts of the endeavor. 2.5 Setting Quality Metrics and Milestones The implementation of life expectancy models can be organized and managed just like any other project. The planning phase in Task 2 produces a list of desired tools and applications and the durations and resource requirements for their development, which can be estimated. These can be sequenced on a Gantt chart as in Figure 2-7. After Task 3 is completed, the data require- ments and applications will be understood in much more detail so the Gantt chart can be refined. If delivery is conceptualized as a collection of separate small applications and reports, as rec- ommended in the preceding sections, then progress can be measured by tracking completion of the individual phases of the individual applications. The phases of each application and report are as follows: • Requirements listing and mockup • First prototype • End-user review and comment

plan for Implementation: how to plan Life expectancy Models 19 • Second prototype • Subsequent prototypes if applicable • Final delivery and installation • Documentation • Training if applicable When the work plan consists of small deliverables, it is not necessary to characterize each phase by percent completion because each phase of each deliverable is either complete or not. The total number of completed phases provides the percent completion of the project as a whole. If a delivery does not meet the end-user’s quality expectations, then an additional prototype may need to be added, which reduces the percent completion until the additional prototype is delivered. Following delivery and implementation of the life expectancy models, long-term follow-up is necessary to determine whether the life expectancy predictions are reliable and to make any requested corrections. For long-lived assets such as bridges, it is necessary to break up the life span into condition states or service levels whose duration can be measured in a more reasonable amount of time. The AASHTO Bridge Element Inspection Manual (AASHTO 2010) provides an example for this application. Figure 2-7. Example project schedule for life expectancy tools.