Diagnosis

The second step in the process of safety management is diagnosis. What does this mean?! It means that there is a first step in this process where the safety engineer has tried to first screen the transportation network in order to determine the most dangerous locations where there is at least one safety problem. There are six steps in this process that in the first step, based on one or several criteria different locations (road segments, intersections, etc.) on the network are evaluated and prioritized for further safety investigations. Now in the second step the traffic safety engineer takes care of that further investigation on a select list of locations that are found most hazardous.

In order to have a clear idea of what is going on in those locations, we need to review several different types of data sources related to safety, field conditions, and other supporting documents. The reason to go over all this data is to find some patterns that can lead us to the problem. For example, if there are too many angle crashes at an intersection then we know something has to be done for it, or if there is too many run-off-road crashes at a stretch of highway (which usually includes curves), then there is something wrong with it and some safety measure has to be implemented.

The point is that we know crashes have the nature of accidents. They are random; we don't know when a crash might happen because of some initially unknown reason. That's why we take a look at the crash data in the last years to see if the occurred crashes are distributed randomly, i.e. not aggregated at specific points and scattered through the segment. Also, It is good to know that examining more years of data provides higher reliability for our finding. Commonly, 3-5 years of data is used for this purpose.

Safety Data

The safety data used to diagnose the problem usually includes the past years' crash data and then see what types of crashes happened more, or whether the crashes are more severe or less, or what days of the week or time of day the crashes happened more, comparatively. This is done by mapping the crashes using GIS, that means putting some dots on the map based on the location where the crashes occurred so we can overview how the spatial crash distribution looks like.

Sometimes, a simple spreadsheet work of crash data handling can be used to diagnose the crash patterns. Simply by creating graphs and histograms the density of specific types of collisions or crash severity levels can be detected. For example, a histogram of number of crashes over a length of the highway (usually the study area) shows that where in the highway was the most crash prone location. Here is a list of example crash graphs that can be generated for the purpose of diagnosing the problems.

The graph presented below is an example. This graph shows that for example, there has been 35 crashes happened only between the log miles 1 and 1.1 in the study roadway. This data can be for 3 to 5 to any number of years in the past crash history of that roadway. Or if our interest is the rear-end collision type which usually happens at intersections, we may conclude that there is higher number of this collisions in mileages 1-1.1 and 2-2.1. I have created a file in which I presented a sample of different charts and graphs for this matter. It can be found here.

One might ask where is this data coming from. 

We all know when a person is involved in an accident, they call 911, right? The police officer comes and takes the driving license and registration documents of the parties involved in the accident and goes back to his car. There s/he enters all different kinds of information into a crash form such as person's demographic information, roadway information, weather condition information, vehicles' information, type of crash information, almost everything about the accident, and a report of how it happened along with a narrative description of the parties involved.

All these crash forms are then taken to the police department and inputted into computer software, e.g. excel spread sheets with all the properties for each crash identified with a number. At the end of each year, department of transportation gets a copy of these crash data and starts analyzing them for safety issues and planning purposes. That's when we are dealing with HSM and following all these steps with the crash data acquired from the police department. 

Supporting documents

Documents acquired from local agencies such as local transportation agencies can help identify the problems. For example, the recent roadway widening projects that have been conducted in that stretch of highway may lead us to know why the traffic volume has been increasing since the time of that project, which has resulted in so and so crashes.

Field conditions

The goal of gathering field condition is to document the typical experience of a road user by actually experiencing the roadway or the other features of the roadway such as bike lanes, left-turn lanes, pavement condition, landscape, lighting, U-turn lanes, intersections, and many more. This data can validate the results found from the crash data or supporting data. We should know there the field investigation done here as part of the diagnosis step in the safety management process is similar to but different than a Road Safety Audit (RSA). For information on what a RSA is and how it is conduced click here.

Now that we know what is wrong with our roadway segment, intersection, or whatever the unit of our safety analysis is, what is next? Probably we are interested in knowing how to solve the problem, right? We'll have to know a little about the crash countermeasure which is the next step in the safety management process.