4.3 Traditional Approaches to Solving Physics Problems

figure4-4

Figure 4: real-world crash test [20]

4.3.1 Real World Modeling

The first most commonly used solution to physics problems is real world modeling. For example, the Insurance Institute for Highway Safety (IIHS) performs crash tests on almost every vehicle commercially available. They perform these tests in large facility in Ruckersville, VA. Each time they perform a full crash test they use a brand new car, crash test dummies as well as barriers to perform the test [21].

Pros:

  • Incredibly Accurate

Cons:

  • Ineffective Prediction
  • Expensive

Incredibly Accurate:

Real world modeling is very precise. Car accidents happen in the real world; it makes sense for these accidents to be tested in the real world. The IIHS has crash tests including front, side, roof, and head restraint tests. All of these tests are attempting to attain a large amount of accuracy with the tests and ultimately rate cars based on their overall safety [22].

Ineffective prediction:

Real world modeling is not the best at predicting the safety of vehicles. Many issues pervade the crash tests because only a finite amount of them can be performed. For instance, the IIHS side impact test uses a “vehicle” at a regular car height crashing into the side of the test car at 31 mph. However, it is just as likely that a car will hit the side of another car at much faster than 31 mph, or that the cars will not hit directly in the center. The IIHS does not test these possible outcomes and so their ramifications are not known [23][24].

Expensive:

Real crash tests are very expensive when compared to other methods. According to Kelley Blue Book the average new car costs $33,543. Each time one of these cars is tested that new car cost is forfeited. Even ultra rare cars such as the Koenigsegg Agera R worth a staggering $1,611,000 must undergo the same crash tests as every other car [25][26].

4.3.2 Modern Computer Simulation

Another commonly used tactic is modern computer simulation. This can be done on nearly any computer so it is very accessible. Most scientific physics engines are created by the group that is conducting the research so there are not many commercially available however that does not stop some scientific groups from performing quantum simulations to test the rules of our physical world [27].

Pros:

  • Cheap
  • Fast
  • Profound

Cons:

  • Inaccurate
  • Difficult to implement

Cheap:

A nice desktop PC can be bought for around $600 in today’s market [28]. This is the only barrier when attempting to create a physics engine. After this cost, only the cost of electricity and internet are considered which are extremely cheap when compared to the cost of $33,000 on average for each real life crash safety test.

Fast:

Simulation is very fast when compared to real life testing alternatives. Honda Motor Co. has adopted a strategy of using simulation to test new car models and determine their safety before they are officially tested [29]. A company named 3DEXCITE has created very accurate crash testing software for Honda to use in their design process.

Profound:

Computer simulations can be used to find more information than what is just on the surface. Honda Motor Co. is using their new software to examine specific structural deficiencies in their car designs. They are able to take apart and put back together a car in a matter of seconds and exactly determine what parts are failing in the car [29]. Apparently, this is massively useful information when designing a car and cannot be achieved using real world modeling.

Inaccurate:

Virtual simulation is very accurate today but not as accurate as real life modeling. Honda is hoping to achieve 100% accuracy in its tests [29]. They have not been able to achieve that accuracy and it is highly unlikely that they ever will with conventional computer methods. This is because there are always a finite set of data points in a computer simulation whereas in the real world the amount of data is infinite.

Difficult to Implement:

Computer simulations are harder to implement than their real world counterparts. For example, a simple physics demonstration would be throwing a ball to the ground and seeing how many times it would bounce in the air before coming to rest on the ground. In the real world this is very easy to implement, there are 3 steps. Get a ball, then throw it on the ground; lastly, count the number of bounces. In order to recreate this task in a virtual environment the programmer must not only know how to program but they also must know basic force concepts, conservation of energy concepts, they must understand the coefficient of restitution and they must understand how to program graphical entities. Obviously, just throwing a ball on the ground is much easier.

4.3.3 Continuing Examples

Aerodynamics

Modeling aerodynamic properties in the real world is incredibly expensive and often cannot test for all possible types of turbulence. Conventionally simulating aerodynamics misses many details that a real-world simulation would not miss.

Soft-Body Physics

Soft body physics fall victim to the same negatives present in both real-world simulation and computer simulation. Trying to test a car for crash test ratings in the real world is incredibly expensive even though it is accurate in today’s world. Trying to test car crashes in a regular physics engine would not be nearly accurate enough to meet the high safety standards of IIHS (Insurance Institute for Highway Safety).

Climate Science

Climate science often uses a mix of real-world simulation and computer simulation to predict the condition of the global climate. As most people will tell you, the weather forecast is not always very accurate. There must be a better way to model these problems and solve the issues they present. The answer, of course, is Big Data Science.