Remembering the case of the failed gas tanks.

Click Image to Enlarge

This is a plot of elastic modulus as a function of temperature for an as-molded PE sample and one that has absorbed gasoline. The effect of the gasoline absorption is evident. The effect of elevated temperature is also apparent in these plots and accounts for the fact that most of the field failures were observed in areas where ambient temperatures were elevated. Testing at 60 C became an important tool in accelerating the failures and determining which designs were most susceptible to performance problems

Over time, the small-gas-tank industry had standardized on an HDPE with a nominal melt flow rate of 4.0 g/10 min and a nominal density of 0.946 g/cm3. These tanks were not subject to the same emissions regulations as those that are in place for the automotive industry and therefore there was no requirement for the multi-layer constructions that are typical for fuel tanks on larger vehicles. 

It is fairly well understood that polyethylene and gasoline are chemically similar and studies have shown that at saturation HDPE can absorb as much as 8-9% gasoline by weight, an amount similar to the saturation level of water in nylon. Perhaps less well known was the effect that this absorption has on the properties of the polymer, because no one had ever encountered a performance problem—until the manufacturer of this grade of material decided to discontinue the grade. 

At that time there were few commercial options in the same density range. Those that were available had lower melt flow rates (MFR), in the range of 1-2 g/10 min. While these would have been viable options, the phobia regarding lower MFR within the processing community kicked in, and instead a material with the same MFR provided by another material supplier was selected. But it did not have the same density; the new offering had a nominal density of 0.952 g/cm3. Within eight to nine months after the changeover, cracks began to appear in gas tanks that were in the field. They all appeared in consistent locations for a particular design and examination of the fractures showed that the cause was environmental stress cracking (ESC).

ESC is the most common cause of field failures in products made from polymers and it continues to be poorly understood by many designers and engineers. Because chemical exposure is always a component of ESC, it is often mistaken for chemical attack. However, ESC requires exposure to both stress and a chemical agent. Without the stress component the chemical is harmless and without the chemical the stress is manageable. ESC is fundamentally a creep mechanism that is accelerated by a chemical, it is not a chemical compatibility issue. In the case of these tanks much of the stress comes from the usual suspects. Molded-in stresses arise from differential cooling rates caused by changes in the wall thickness of the part or poor cooling around cores and hot tips. 

Orientation effects can also contribute. In addition, the assembly process for these parts creates another factor. These products are molded in two pieces and then thermally welded together within a few minutes after they are ejected from the tool. If the parts are allowed to cool completely they become too warped to assemble. If the parts are welded before the cooling process is complete, then the additional shrinkage that occurs in the assembly takes place under constraints provided by the weld. It was not a coincidence that most of the failures occurred near the weld and almost always appeared in the lower half of the assembly where gasoline exposure was more of a constant.

The importance of design was illustrated by the fact that not all tank designs exhibited failures. One manufacturer had 19 different tanks, and only four of them displayed failures under controlled test conditions that combined gasoline exposure with elevated temperatures. If the problem were chemical attack, they would have all failed. 

Figure 1 shows a plot of elastic modulus as a function of temperature for an as-molded PE sample and one that has absorbed gasoline. The effect of the gas absorption is evident. The effect of elevated temperature is also apparent in these plots and accounts for the fact that most of the field failures were observed in areas where ambient temperatures were elevated. Testing at 60 C became an important tool in accelerating the failures and determining which designs were most susceptible to performance problems.

While all of this was useful in helping to understand the problem, it is important to note that none of the tank designs had changed and there were no unusual climatic conditions that occurred in 2001 that had not been around in previous years. The factor that had changed was the raw material. Resistance to ESC is one of those properties that improves as molecular weight increases and density decreases. In most polymers we focus on molecular weight because density is not a variable. 

But in PE it is an important variable and it must be considered as part of the performance equation. The lower density material was not impervious to the effects of elevated temperature and gasoline absorption. But the lower level of crystallinity in the 0.946 material provided a structure with greater ductility for a given molecular weight, allowing the parts to elongate to higher strain levels without breaking. The higher density material was too well organized at a molecular level and responded by cracking rather than stretching. 

This runs counter to the relationship between crystallinity and ESC resistance in most polymers. We are taught that ESC primarily occurs in amorphous polymers and crystallinity is a major deterrent to the mechanism. By this logic a higher density PE, with the corresponding higher degree of crystallinity, should work better. But in practice the opposite is true unless we replace some of the lost ductility by increasing the molecular weight at the same time. In applications where ESC failures are simply intolerable, the material specification reflects this leveraging of the combined benefits of high molecular weight and lower density. Liners for natural gas cylinders are made from materials with densities in the 0.937-0.944 range and with melt flow rates of less than 0.1 g/10 min when measured under standard conditions. 

The failure to even try the higher molecular weight materials that were still being made by the same supplier who had discontinued the incumbent material and instead change to a higher density option unleashed a crisis that affected everyone using these smaller gas tanks. The problem was resolved when a third PE supplier brought out a 0.945 density grade with a melt flow rate of 3.3 g/10 min, close enough to 4 to keep from scaring the molders. 

But the damage was done. The Consumer Product Safety Commission got involved and the fallout has resulted in new rules and regulations on tank construction that many undoubtedly believe are needless and costly. Still, government regulation is a bell that is nearly impossible to unring, even when good science offers a less cumbersome solution. Maybe  this was a lesson learned and the knowledge gained will serve users of PE in the future. But maybe not. In China recently Type IV gas cylinders experienced spectacular explosive failures that have been diagnosed as being caused by ESC in the PE liners. I wonder what grade of PE they used.