Carl F. Thelin

FAILURE IS AN OPTION IN R & D

   

         That is what I told young engineers designing something new and different. Do you expect the word research to mean only one kind of activity, such as doing scientific experiments in the physical or chemical fields? R &D stands for Research and Development. Research involves two activities. Do your “re-search”. What is that hyphen doing there? You should make that re-hyphen-search effort first. Search the records to find out what others have done. Make your own list of all those other ideas that did not succeed. Learn of their failures so you can eliminate them from the research experiments that you plan.

            EXPERIMENTS

            Making your own mistakes - failed attempts - is experimental research. Plan your own experiments. It is OK to make two mistakes alike – maybe you skipped a step in the first experiment. Do not try it the third time – make a new mistake .You should understand that some ideas do not work out. Call that idea a mistake. However, you are dumb if you make the same mistake three times.  When you make all the mistakes there are to make, what is left is the right thing to do. You should also know that it is more acceptable when you do the failed experiments by calculations on paper or a computer screen; not in hardware.

            Development is what we do once we settle on a feasible idea – one that works, is producible and is affordable. I grew up in organizations that were designated "R & D." Research and Development, whether for academe, industry, or government has facets of importance to us. General Motors, CAL and the NHTSA had large, expensive facilities and highly skilled people to do experimental research with cars and trucks. A physicist in those laboratories sought new information by experimenting with tools so expensive that few individuals could afford them. Places with big labs also had other researchers -- the ones with the hyphen: persons who searched libraries for references to the research done by other individuals and groups. There is no point in doing something twice, if the first time was good enough. Sometimes you could combine the findings of several researchers to reach a new conclusion. Library research is the second kind of research.

           

            MODELING

            Now there is another class of researcher. This one combines experimental skills with knowledge from other sources to develop new understandings. This person was once as rare as Albert Einstein or Steven Hawking, who created elaborate models of the universe in their minds. Now we create accurate imaginary models with which to experiment in a laboratory that sits on the desk. The personal computer permits us to simulate events that we cannot otherwise examine. Realistic modeling of probable scenarios is the third kind of research.

            Here is a now-common way modeling is used. There is a type of highway collision that we call the "near-hit -- near-miss" scenario. An example is when the driver of a car waiting at the stop sign of a cross street pulls onto the highway while turning left just in time to be hit by a truck coming from his left. At CSI, we gathered as much information about the collision as was available. Then we experimented with scientifically validated computer programs. These allowed us to examine many possible outcomes given a range of assumptions in association with the known facts. We experimented with variations in the appropriate speed of the vehicles, or with the frictional nature of the surfaces, or with the timing of the truck driver's reaction to the cross traffic or the car driver's response to visible threats. These computer simulations are dependent upon the skill of a teammate who gathers the physical data like the length and character of the tire marks. The computer model also needs data found in various databases for numbers such as the mass and crash stiffness of vehicles. The personal experience of the researcher counts, too. He must apply his judgment and chose the most probable values of variables in the model.

            The role of the expert accident reconstructionist is to examine all the available evidence. Some eye witness descriptions will agree; others will not. The expert combines all the data and searches for the scenario that has the least contradictions. The ability to examine quickly and cheaply many combinations of distances, timing, and sight lines enables the experienced reconstructionist to arrive at a reasonable conclusion and illustrate it so convincingly that others will share it. Using computers to produce realistic three dimensional visual images is the latest enhancement. Now the expert can use graphical output to compare each result of his experiments. Then he can display his conclusion pictorially so that even a layperson will agree that is what most likely happened.

            Back to the example

            The CSI needed experimental data. He measured the rate at which the truck would coast down if the driver were to lift this foot from the throttle pedal. This would vary with the grade of the road as well as with the engine size and transmission ratio. He had the means of getting a continuous plot of speed and location of the automobile as it accelerated and turned left, depending upon the radius of the turn and the power of the car's engine. The investigator must use some judgments, since it is unlikely that anyone would make such a maneuver at full throttle. Then he must find appropriate values for the coefficient of friction, if the truck driver was able to get onto the brakes before the collision. If the collision occurred at night, he must consider the effects of the truck's headlights. If the lights were on high beam, the additional glare might prevent the driver of the car from seeing the truck's amber clearance lights. The cab-top clearance lights should tell a driver that the oncoming headlight spacing are narrow because they are on a truck.  Otherwise, narrow spacing of the head light means a car is far away.

            The expert stirred all these things into the mixture for examination and simulation. If he calculated that the driver of the truck was traveling well above the speed limit, he will ask this question: If the driver of the truck had been going at the speed limit, would the car have safely completed it turn before the truck reached the point of the collision? Otherwise, he may find that the driver of the car waited too long, and pulled into the path of the truck when there was no time remaining for the truck driver to avoid the collision.

            The synergism of our large forensic engineering firm enables it to utilize all three kinds of research to develop sound analyses and clear presentations for an investigation.