A forensic hypothetical: The case of the percolating water

By Dov Leshchinsky

Introduction

Nearly every SRW system in the market has experienced some sort of failure. Like the power of chum in luring sharks, failures attract attorneys and engineering experts.

Typically, the attorney for the plaintiff will compile “expert” reports providing a list of what appear to be the faults in design and/or construction. Generally, most of the listed items may be irrelevant to the actual failure but could represent potential generic problems. While such a long list clouds the actual reasons for failure, it has the potential to impress a jury or a judge who likely understand very little regarding technical concerns related to the SRW system. After all, the fact is that there was a failure. Consequently, the judgment could be a hefty “compensation” for what appears to be an engineering problem.

To avoid the risk of facing such a situation, and regardless of the engineering facts, the attorneys for the defendants will seriously consider a settlement whereby the financial outcome is potentially smaller than the cost of a good defense or the settlement from what an unpredictable judicial system may decide. It is a simple matter of business, and the end result may not entail justice, as the likely truth could be sacrificed for a compromise/settlement.

Unfortunately, the lesson that should have been learned from a failure, which is so important for advancing safe engineering practice, is blurred. This, no doubt, results in needless future failures of similar structures while creating more chum for attorneys and experts. Can we guess who the real winners are: the designers, the contractors, future owners, or the attorneys and their experts?

The author has served as an expert in several litigations, a few including SRW. Though it is well known that SRW may fail due to seeping water, it is felt that an instructive and simple forensic study of a case should be demonstrated.

This study shows that some irrelevant aspects are frequently attributed to failure while the major culprit is overshadowed or underplayed. It is hoped that this article will reduce the number of failures in SRW systems associated with water. For legal reasons, the described case is hypothetical; however, it is likely that most of the existing SRW systems have experienced a similar collapse or nearcollapse situation. Hence, the hypothetical case is both generic and realistic.

The problem

Forensic study is a process of reverse engineering; we know that the factor of safety for at least one potential failure mode became less than 1. That is, while the end result is known, there could be several plausible reasons for it. The challenge is to identity the most probable reason or combination of reasons.

Ideally, one would explicitly present the anatomy of a specific failure. However, because of the legal system, this is a luxury that can rarely be afforded. Hence, we illustrate a hypothetical case that combines elements from several experiences to yield an instructive lesson. Figure 1 is a not-so-funny, albeit realistic, cartoon of a failed SRW. This image depicts a signature failure of percolating water at the crest.

Assume that a lawsuit ensues and the plaintiff claims that the culprits are erroneous connection strength used by the designer, improper construction, defective blocks, and whatever else can maximize the number of deep pockets to drag into the case. Few facts need be provided by the plaintiff, just allegations. To dismiss the potential impact of water, the plaintiff claims that, indeed, there was record rainfall but it actually happened a few days prior to the collapse, thus, it is not relevant.

The plaintiff says the backfill is sand, forgetting to point out that it contains more than 20% fines, implying that it is freedraining, meaning that water will percolate quickly through it. Besides, the plaintiff argues, the block facing cannot retain water, and therefore there is no hydrostatic pressure exerted on the back of the wall. Hence, the plaintiff portrays the rain event as important only for making green grass greener. Armed with photos of the failure, the plaintiff needs little science or engineering to make a convincing case in front of a jury. Failure most certainly is the result of weak connection, poor construction, defective materials—all the necessary ingredients to explain the cause of failure to a layperson.

The problem? The defense needs to identify the actual culprit, using objective and sound engineering tools.

Forensic study

Poor defendants: May the power be with their attorneys and experts trying to explain to the court the likely reasons for failure actually using engineering principles. Conveniently, this article concentrates on the forensic expert study so that the judicial defense problems are left to the attorneys. No doubt, it is easier on the soul to be an engineer. (But this may also be reflected in the respective compensations.)

Following accepted design procedures (e.g., AASHTO, NCMA), the defense expert should check the reinforced SRW system for stability. Design procedures are conservative, as they should be, and therefore, such a check is useful in assessing the robustness of the system. Typically, the resistance aspects in design calculations are underestimated while the loading elements are overestimated. Consequently, a redundant system is produced, and failure of a properly designed system requires the unusual occurrence of several events simultaneously.

Using the actual design calculations as a forensic tool by the plaintiff is useful, as any deviation from accepted industry practice will complicate the defense. However, for the defense, it may not explicitly identify the major reason for failure, but it demonstrates potentially poor engineering practice. Furthermore, accepted design methods are based on lateral earth pressures, which are lousy predictors of actual loads, especially at the connection of flexible structures such as reinforced SRW. Most importantly, while redundancy exists for the assumed field conditions in design calculations, there is none when water unexpectedly seeps into the system. The assumption that no water pressure can develop behind stacked blocks, especially when a 1-ft gravel “drainage” layer buffers the backfill and the back of the blocks, leads to false confidence, as shown later.

Checking the robustness of the system using an existing design method, let us assume that the expert for the defense finds that the margin of safety against long-term connection rupture strength at the bottom of the wall is low, but greater than unity. Could this be a reason for failure? Looking at Figure 1, it is apparent that blocks in the upper portion of the wall collapsed, not at the bottom as implied by the design-related analysis. The margin of safety for rupture and pullout connection strength in the upper portion of the wall is adequate. Besides, the failure occurred within the first year after completing the construction; long-term based on design refers to 2 orders of time magnitude greater when the connection capacity would be nearly half of what it was at the time of failure.

Though the appearance of falling blocks tempts one to claim inadequate connection as the culprit, conservative reassessment of the failed SRW system shows that it is an unlikely reason. At this stage, use of existing design helps the defense to exclude an important argument by the plaintiff. The defense still may need to explain why the connection at the bottom is not as strong as required by the design procedure. However, this is irrelevant to the failure.

It is not unusual to find more fines in granular backfill than specified in design. In fact, design methods such as NCMA are very liberal in the allowed amount of fines.

Back to our case, regardless of whether specified or not, assume that the defense identifies the backfill as sandy with about 20% of particles smaller than 0.075mm (i.e., fines). Such quantities of fines slow the flow of water should it somehow seep into the reinforced soil zone. The seepage velocity could be less than 1m per day, implying that water can accumulate and saturate some of the backfill, especially if fed through a defective surface or subsurface drainage system. Subsequently, the argument by the plaintiff that the heavy rainfall is not relevant as it occurred a few days before failure is not necessarily relevant. It may take a few days for water to move through this “free-draining” reinforced soil.

Whether defective construction materials were used or not can be verified. However, the local nature of failure implies that such defective materials (e.g., blocks) will be placed in a very limited zone, not a likely situation. Whether construction was poor (e.g., poor compaction or placement of blocks) or adequate can be verified by field tests on stable portions of the wall and construction records. The localized nature of failure again indicates that the inherent redundancy of the system had to be severely violated, thus making it unlikely that construction is the main culprit.

It is time now to assess the impact of possible water percolation. Can it cause a failure resembling the one observed? Is there a large volume of water needed to create such a failure? Can it be the culprit?

Impact of water percolation

Assume that in our case, failure occurred next to a driveway, which directed surface water runoff toward the reinforced soil zone. Hence, localized source of water, especially after rainfall, is feasible, thus making the study of water percolation necessary.

The internal stability of an SRW system is derived from the connection strength, the geogrid strength, and the soil shear strength. The contribution of small blocks to the stability of any sizable wall is small and can be ignored. The connection strength is virtually independent of water, especially when it is not truly submerged. The long-term geogrid strength is also independent of water.

However, the strength of the soil depends on its intergranular friction and the intergranular normal force or normal stress. The friction is virtually independent of water. However, submergence reduces the effective weight of the soil by as much as 50%. Consequently, water pressures can decrease substantially the inherent soil shear strength.

In such a case it may not provide enough resistance to maintain the reinforced soil mass in static equilibrium. Reduced soil strength may also limit the pullout resistance of the geogrid, in its front or rear end, thus limiting its capacity to contribute to stability. Hence, hydrostatic water pressure against the facing is not the only potential impact percolating water. The significant reduction of soil strength is enough to render an unstable SRW system.

The impact of water on stability through the strength of the soil can be assessed using slope stability analysis. For our problem, circular potential slip surfaces are considered using Bishop’s limit equilibrium method. Program ReSSA(2.0)¹ , which is particularly convenient for SRW since it considers connection strength in equilibrium, is utilized. ReSSA has the safety map feature. It is a diagnostic tool signifying the safety factor at each location within the soil mass considering large numbers of circles passing through that point.

Figure 2 shows the safety map for the dry case. The minimum safety factor, i.e., the factor of safety for this problem, is Fs = 1.45. It should be noted that the reciprocal of the factor of safety in limit equilibrium analysis signifies the level of mobilization of the soil shear strength. For example, a safety factor of 1.5 means that on average (1/1.5 = 0.67), 67% of the soil shear strength is mobilized at that point of interest. The zone within which the safety factor is between 1.45 and 1.50 is signified by the red color in Figure 2. It is a very narrow zone. Clearly, when dry, the system is stable using global stability approach.

Figure 3 shows the safety map when the entire reinforced soil is saturated. As can be seen, the zone within which the safety factor is less than 1.0 is rather narrow, all located above the bottom grade elevation. Safety factors less than 1.0 imply that the system will collapse in that zone with the facing units falling. As can be seen, the “red zone” produce a signature failure related to water resembling the one seen in Figure 1. Comparing Figures 2 and 3, one realizes that the presence of water approximately cuts in half the dry factor of safety, a very significant impact.

Furthermore, the red zone implies that a massive saturation is not needed; that is, it does not need to extend all the way to the bottom and it does not have to be far away from the back of the blocks. Since soil is composed of solid particles and voids, and since the volume of voids needs to be filled with water for saturation, one can estimate the amount of water needed to percolate in order to reach saturation of the red zone and create failure.

Take the porosity [porosity = (volume of voids/bulk volume of soil)] as a typical value of approximately 0.5 and multiply it by the area of the red zone to obtain the required volume of water for saturation per unit length of the wall. For the case in Figure 3, it is less than 10ft³ (62 gallons) per ft-length of wall. As can be seen in Figure 1, the length of failure is rather short, meaning the total amount of water needed to generate a signature failure can be easily produced by a rainstorm if the drainage system does not prevent water percolation.

It should be pointed out that at failure, the geogrid need not rupture. It can stretch a little with the connection breaking (or pulling out), the soil sliding over the geogrids at their front end, as a rather shallow wedge is formed within the reinforced soil (see Figure 1). This mechanism may be realized as part of the analysis.

While calculations were done using the long-term geogrid and connection strengths, it is likely that the available strengths contributing to stability are larger, as the increase load due to percolating water is relatively quick (a few days). It can be shown that this may render somewhat larger safety factors. However, these would generally not be significantly larger. If pullout controls the connection strength, its long-term strength is the same as short-term, and in this case there would not be any effect on the safety factors. In-depth discussion of time effects is beyond the scope of this article, but it will not change the lesson learned about the potential impact of even a small amount of percolating water.

Conclusions

1. Design methods that are based on lateral earth-pressure design (e.g., NCMA, AASHTO) should render safe reinforced SRW systems. The inherent conservatism (or redundancy) in these design methods makes them less desirable for forensic studies. Conversely, when water is a suspect, forensic analysis based on dry conditions using these methods may overpredict stability leading to unrealistic conclusions.
2. Slope stability analytical approaches yield reasonable results when failure due to water is suspect:
   • Stability is high when dry conditions prevail. This corresponds to a reality in which no percolation occurs.
   • Percolating water has significant effects on stability. The volume of percolating water or its extent need not be large. Heavy rainfall combined with poor surface drainage can produce sufficient water for local critical percolation. Excessive fines within the reinforced soil eliminate the free-draining properties of the backfill, thus facilitating slow flow leading to possible saturation.
   • The predicted collapsing mass, starting at the crest and extending to some height of the wall (i.e., not necessarily to the toe), agrees well with the field observations of failure. The author considers the collapse in Figure 1 as a signature failure due to water percolation.
3. The so-called “drainage layer” of 1ft gravel may not be effective in preventing failure due to water percolation. If the reinforced soil gets saturated, failure is likely to occur, perhaps in a progressive (time dependent) manner.
4. Percolating water is a frequent cause for SRW failure. Consequently, proper surface drainage or diversion of surface runoff should prevent such failures. It is a detail that should be part of the design and construction. Unfortunately, these considerations are frequently beyond the control of the designer in their actual implementation.

Costs of failure are always larger than the costs for properly designed and constructed walls. Forensic studies are extremely important in educating designers and contractors of proper practice. It is hoped that the realization of the impact of water on SRW stability will help reduce future failures.

Dov Leshchinsky, Ph.D., is a professor of civil and environmental engineering at the University of Delaware and a regular contributor to Geosynthetics magazine. His last article, “Preparing for the LRFD mandate,” appeared in the August/September 2006 issue.

References

¹Program ReSSA(2.0) for stability analysis of reinforced slopes and walls (Version 3.0 is now available): www.GeoPrograms.com.