Dana W. Clarke, Sr.,
Ideation International Inc.
Most designers are at least somewhat familiar with Failure Modes and Effects Analysis (FMEA) and Hazard and Operations Analysis (HAZOP). These methods attack problems by organizing the available information in ways that sometimes let solutions fall out by inspection. FMEA, for example, forces developers to categorize failure modes, put a number on the degree of failure risk, document actions taken, and perform other tasks aimed at quantifying the problem.
Existing failure-analysis methods have one drawback, however, in that they don't offer structured methods for actually finding or anticipating root causes. That is one reason why a relatively new technique called Anticipatory Failure Determination (AFD) can be helpful when dealing with failure modes.
AFD differs from older failure-analysis techniques in several ways. But perhaps the most strategic difference is how it leads development personnel to examine failure modes. AFD turns a hardware or software failure on its head. Rather than asking developers to look for a cause, AFD has them view the failure of interest as an intended consequence. Developers then try to devise ways of ensuring that the failure always happens and happens reliably. Invariably the real cause of the problem becomes obvious from the results of these studies.
In contrast, ordinary failure-analysis methods generally view failure modes in a conventional way, asking participants to theorize what might be happening. This viewpoint may indeed get participants to a solution, but perhaps less quickly or directly than focusing on how to make the failure happen. Moreover, conventional failure analysis relies on unstructured procedures such as brainstorming to come up with possible causes. AFD, an off-shoot of a systematic technique called TRIZ, employs well-documented procedures that make technical innovation a rigorous process.
AFD is a far more general failure analysis tool than HAZOP or FMEA, because its basic methods are not based on any particular technology or industry. The methods can be used to analyze any type of process, system, or even soft issues in an organization.
Because of its general nature, the AFD process and the steps it takes to a solution can sound extremely abstract and hard to grasp initially. That's why the technique might best be explained through an example problem solved by application of its principles.
FORMULATE THE PROBLEM In one case, 150-mm shells were exploding immediately after leaving the gun barrel. This was puzzling because 120-mm shells of the same design worked perfectly. The first step in the AFD process is to formulate the problem and spell out the original situation. Although this sounds like common sense, studies indicate that many engineering teams don't take the time to do it rigorously. It takes place casually, is never committed to paper (or a database), or never gets farther than conversations among team members.
For the prematurely exploding shells, the unwanted effect was premature explosion with the shell just leaving the gun barrel.
IDENTIFY "SUCCESS" The next step in AFD asks for a spelling out of the success scenario. Engineers studying the problem divided events into four phases: a shell first pushed by gunpowder, then moved out of a gun barrel, next moving toward a target, and finally hitting a target. The required result is that the shell must stay in one piece during the first three phases, then explode during the last one.
LOCALIZE THE FAILURE Tests showed that things proceeded apace until phase three, when the shells would often explode prematurely. Thus troubleshooters could deduce that something was going wrong before or during phase three.
FORMULATE AN INVERTED PROBLEM AND AMPLIFY IT It is here that the problem gets turned on its head. The inverted problem statements would read something like, "Produce premature shell explosion before or during movement toward the target (phase 3)." The amplified problem statement: Guarantee the shell will explode prematurely immediately after it leaves the gun barrel. The amplified problem statement makes the situation extremely dramatic. This is done intentionally to help overcome psychological inertia and ensure the ultimate solution is reliable.
SEARCH FOR SOLUTIONS When the shell round actually hits its target, inertia forces compress a spring attached to a firing pin. The pin then hits a blasting cap which activates the explosive. Thus, premature explosion requires some force capable of pushing the firing pin toward the blasting cap without the shell ever hitting the target.
The AFD method next categorizes the resources available to provide the required effects, in this case, the force pushing the firing pin. More than just a general list, this categorization divides resources into seven areas: substances, field effects, space available, time, object structure, system functions, and other data on the system. This listing helps ensure that potentially important factors aren't ignored during analysis.
The substances operative in the shells included the steel barrel bottom of the shell, the steel spring, and the explosive. Field effects included inertia and the pressure of powder gases inside the gun. The available space consisted of the volume under or around the detonator. The time period of interest was that before and immediately after the gun shot.
The next step in AFD is to examine the available resources to see how they can play a part in providing the outcome wanted. This can take place by hand, but AFD can also employ TRIZ software that provides some guidance here. For the artillery shell, it pointed toward examining the forces present. Detonator inertial force was in the opposite direction of that needed to cause detonation. Powder-gas pressure was in the required direction, but didn't influence the firing pin, at least not in the smaller 120-mm shell.
The flawless operation of the smaller shell was a contradiction of sorts. The AFD method teaches a structured approach to resolving contradictions. Here, it advised looking at the space resources of the two shells and how they differed, because dimensions were the only real difference between the two cases. In particular, the 150-mm shell had a barrel bottom that was larger, though its thickness remained the same. Changing the diameter of the thin metal shell bottom without increasing its thickness made it more flexible.
FORMULATE HYPOTHESES AND TESTS THAT WILL VERIFY THEM This analysis led to a hypothesis that powder-gas pressure caused the thin shell bottom to distort inside the shell. The detonator, fixed to the bottom, moved with it. When the shell left the barrel, powder-gas pressure dropped quickly and the force of elasticity returned the bottom to its normal position. But the firing pin, not fixed to the bottom, continued its movement forward and hit the blasting cap. Tests verified that this indeed was the case.
All in all, following the AFD technique can quickly get trouble shoot-ers to find root causes that might otherwise be hard to envision. There is commercially available software that can help in determining the pertinent resources and promising avenues of pursuit. But even without the aid of software, the AFD approach can help organize efforts to quickly reach a fruitful conclusion.
THE STEPS IN AFD
- Formulate the original problem.
- Identify the "success" scenario.
- Localize the failure.
- Formulate and amplify the inverted problem.
- Search for solutions Spell out available resources.
Use them to produce the wanted effect. Look for contradictions.
- Formulate hypotheses and tests for verifying them.
FOR MORE INFO
Ideation International Inc. provides software and training on Triz and AFD techniques.
Site dedicated to improving the FMEA process through consulting services, education, training, and software.
A useful-links page maintained in conjunction with a course on safety-critical systems development taught at the University of Glasgow.