What is the Shelf Life of Data?
Fall is one of my favorite times of the year. The kids go back to school and the bounty of foods from our nation's farms come in from the fields. It also means report card performance and attention to freshness. When it comes to mining data of a dynamic world what is the shelf life?
The latest report card (2013) for America’s Infrastructure from the American Society of Civil Engineers (ASCE) has given our infrastructure an overall D+ (poor at risk conditions below standard with a strong risk of failure). ASCE estimates a total needed investment of $3.6 trillion by 2020 to restore the nation’s infrastructure. Poor road surfaces cost the average U.S driver about $324 per year in vehicle repairs, or a total of $67 billion. Also, it has been estimated that small improvements in road surface conditions can decrease fuel consumption between 1.8 percent and 4.7 percent, according to the American Association of State Highway and Transportation Officials (AASHTO) based on existing conditions. Because our nation’s roads are dynamic, having current as-built data is crucial.
Many factors affect our road conditions. Pavement fatigue include load, materials and surface abrasion due to weather. Ongoing construction projects like milling, leveling and resurfacing add to the dynamics. We cannot possibly accurately represent current conditions using canned data. As a land surveyor for almost 40 years I have learned how imperative it is for design engineers to have accurate current information for bridge clearances, signage, analytics and roadway elevations, especially within an aging infrastructure. The land surveying profession promotes accurate documentation of your property for something as easy as properly adding a fence, regardless of the last survey. So why would we design and build major transportation infrastructure with old, possible changed information?
Obsoleteness of data is an unknown until proven otherwise. Using data beyond its useful shelf life will not give you any valuable insights and poses cost and compliance risks. This discussion is not so much about affordability of capturing and storing information, but it is about making the right decisions to use existing data or obtain new current data to minimize risks, maximize value. This discussion is more about what is the shelf life of highway data. The elusive answer is… depends. Some data may remain relatively current for decades while other data may be compromised in weeks. The concern is which areas?
The overall cost today with Terrestrial Mobile LiDAR Scanning (TMLS) is minimal compared to designing from outdated data. As a surveyor I have always been intrigued by AEC projects that have a budget of tens of millions of dollars but the initial survey budget gets beaten down to the lowest bidder. So, in effect, the start of a design project can be compromised by inaccurate data that is not discovered till the construction phase, then the fix can be many times more dollars to fix than the cost of the survey.
Let’s look at a hypothetical 10-mile-long, four-lane divided highway pavement restoration project. The pavement has been patched, overlaid in some areas etc. What is the cost to obtain a current TIN of the pavement surface: $2,500 per mile, $5,000 per mile or even $10,000 per mile? Let's go to the extreme and use the $10,000-per-mile fee… that would be a $100,000 fee to create an accurate, current TIN. Using a construction cost of $1,000,000 per mile means the updated current survey would cost 1% of the construction cost. Question? What type of overages are seen on a typical project of this type? One percent, 5 percent, it’s something to think about and start a discussion about where the money is spent.