By now you are probably aware of Toyota's accelerator pedal quality problem, which recently prompted Toyota to recall 6 million vehicles and halt production and U.S. sales of 8 popular models. According to stories in the Wall Street Journal, the primary symptom, as reported by Toyota drivers, is sudden, uncontrolled acceleration of the car. It is believed that the problem has two causes; misaligned floor mats and/or sticky accelerator pedals. Toyota's very public and embarrassing situation creates an example for the application (or mis-application) of Critical Parameter Management (CPM), also known as Key Parameter Development and Management (KPD&M) in the aerospace and defense industries. (ed. note: this article was written on Feb 2, 2010)
Discovering and Preventing the Accelerator Problem with CPM
Should it be considered "standard work" to conduct analytical and experimental evaluation of potential sources of variation (noise parameters) that can disrupt the proper functions within and between the automobile's subsystems and other components which are in functional proximity, such as the accelerator subsystem and the floor mat in this case?
From Toyota's own product recall website FAQ section, the following statements help determine if CPM may have prevented this massive problem from such humble parts, assemblies and their interrelationships. All of the underlined terms below are physical functions that are by-products of the design's sensitivity to noise.
Q. What is the problem that could cause accelerators to stick and led to the recall?
A. The issue involves a friction device in the pedal designed to provide the proper "feel" by adding resistance and making the pedal steady and stable. This friction device includes a "shoe" that rubs against an adjoining surface during normal pedal operation. Due to the materials used, wear and environmental conditions (noise parameters), these surfaces may, over time, begin to stick and release instead of operating smoothly. In some cases, friction could increase to a point that the pedal is slow to return to the idle position or, in rare cases, the pedal sticks, leaving the throttle partially open.
If CPM were applied in this situation, the accelerator subsystem design team would have conducted the following five tasks, which would be considered standard work:
1. Generate a Functional Flow Diagram and accompanying Math Models to Monte Carlo Simulate the functions in static and dynamic forms to forecast the effects of variability using the next step's input (Task #2) to the models. They would have computer generated Capability Studies (Cps & Cpks).
2. Generate a Noise Diagram to identify external, part-to-part and deterioration sources of variation that could affect the function of the accelerator. This technique identifies the physics of noise to include in analytical (Task #1) and experimental modeling evaluations (Task #5).
3. Generate a Design FMEA for the main functions of the accelerator subsystem. This method identifies potential failures of functions.
4. Generate a Worst Case Analysis, Root Sum of Squares Analysis or, more appropriately, a Monte Carlo Simulation of the tolerance stack up to explore the boundaries of the designed function of the accelerator's motion, including the intended resistance for pleasing pedal feel and feedback. Such analyses take part-to-part variation into account but have limitations when it comes to wear phenomena and external sources of variation such as thermal or hygroscopic expansion or contraction issues.
5. Conduct Robustness stress experiments to explore and desensitize functions that are sensitive to specifically identified sources of noise. This method evaluates and exploits interactions between controllable engineering parameters and unwanted sources of variation (noise parameters).
When practicing Lean Product Development or Design for Lean Six Sigma, these five tasks are required for every subsystem. It is more likely that Toyota's problems would have been discovered reasonably early in the product development process had a capable CPM process been applied.
The Supplier's Product Development Process Must be Considered
Is it safe to trust a supplier's product development process to properly conduct robustness development tasks that prevent or minimize problems caused by noise parameters such as those encountered in this case? A supplier may claim that all they need are nominal target values and tolerances for dimensions, spatial relationships, shape factors, surface finishes, lubricants, sealants, bulk material properties, and etcetera. They will design to those specifications and meet the customer's requirements "in full". Then when there is a problem, the customer's specifications (critical parameters) are blamed.
What if the supplier does not include robust design and tolerance design as standard work within their product development process and the customer is not aware of it - or worse, does not particularly care?
It's necessary to verify the supplier's product development process details, including CPM processes, methods and results. It's less expensive to verify and then trust the supplier's process. Incomplete supplier management, particularly in the area of CPM, may have contributed to Toyota's problem.
What is the Cost of "Knowing" (or not Knowing)?
In a statement about the recall, Jim Lentz, president of Toyota's U.S. sales arm said,
"We are confident we have identified the causes of the reports of sudden acceleration, which are misaligned floor mats and sticky accelerator pedals. These two fixes solve the issues that we know of."
We now ask, what critical parameter relationships should one "know of" vs. what can be ignored or bypassed during normal product development tasks or standard work? And, what should one know about the functional robustness of a spring-mass-damper (the accelerator) subsystem? Could these sensitivity-to-sources-of-noise issues have been identified by applying CPM? Were these actually unknowable problems that could only be detected after they occurred? Would a standard Design FMEA have missed them? What about the construction of a Noise Diagram? Would it have included the root causes of this problem?
Even if it had cost $1 million, for example, (update: as of February 7, the estimated cost of the Toyota recalls is up to $2 billion) to prevent this problem during development - it would have been worth the critical parameter knowledge about these sensitivities to noise. If it had slowed down the development by 3 months, for example, it still would have been a bargain. Companies may under-invest in CPM and the use of robust design and tolerance design to "stay on time". The cost of proper development of this small accelerator assembly and floor mat is tiny compared to this level of financial loss and damage to Toyota's brand and reputation.
The lesson is hopefully now learned, and it is very likely the design error will not be repeated - unless the people who learned this lesson leave the company and their replacements are not practicing CPM. Then it could happen again.
