28 April 2021
To New Zealand Geotechnical Society members,
Understanding Expansive Soils within the New Zealand Building Code Framework: A Structural Engineer’s View
I am an Auckland structural engineer from Markplan Consulting. Our company has been involved in the design of tens of thousands of raft/waffle slabs on expansive soils, over the past 25 years. Our director, Mark Smith, was heavily involved in raft slab inception in the NZ industry (in the mid-late 1990’s) alongside the likes of Firth. I have worked for Markplan Consulting for the last 12 years and in that time, I have been responsible for developing our design methodologies for raft slabs on expansive soils. As such, I have a good general understanding of the interaction of structural raft slab foundations with expansive soils (by no means an expert in geomechanics).
I have read the recent paper by Tonkin and Taylor (T+T) in the NZ Geomechanics News (June 2020) regarding the shrink-swell test, and found this very interesting. Markplan Consulting carry out many projects with large housing developers in Auckland, and we see many subdivisional geotechnical investigations and reports. The T+T paper helps explain why we are seeing such a large variability in specified soil classes within a single subdivision, and sometimes even within the same building footprint. I am also aware of the further work that T+T have done in this domain and that has been presented at the recent NZGS Symposium. It is concerning that there appears to be significant flaws in the testing methodology, and it seems to me that the New Zealand geotechnical community (and Australia for that matter) has a considerable issue to resolve.
Aside from the issues noted by T+T, I would like to draw your attention to other concerns, which I believe to be causing considerable confusion with structural engineers and with geotechnical engineers, in regards to expansive soils in NZ. These are code/design issues, that do not exactly require a high level of geotechnical expertise, but do require the geotechnical community to assist and progress further industry understanding or change. Some members may already be aware of these issues, and may even be implementing the recommendations that I propose here.
My concerns relate to the November 2019 update to the NZBC, which refers to expansive soils within B1/AS1 amendments to NZS3604. This update is mentioned in the T+T paper, and symposium presentation. Many structural and geotechnical engineers that I have spoken to about the NZBC update are of the understanding that MBIE have made a deliberate decision to revert back to the site soil classes as outlined in the superseded version of AS2870:1996, rather than utilise the Class H split to H1 and H2 outlined in AS2870:2011. I strongly believe that this has been an unconsidered and unintended change. This is based on the following logic:
- AS2870:1996 (now superseded) outlined soil classes and characteristic surface movement (ys values) as Class S (<20mm), Class M (20-40mm), Class H (40-70mm), and Class E (>70mm).
- BRANZ report SR120A (2008), which is mentioned in the T+T paper, references AS2870:1996, which was the active version at the time. One of the main design recommendations of this report was to effectively increase the return periods adopted for foundation design, by way of scaling factors which were outlined in the report.
- In 2010, the Department of Building and Housing (now MBIE), produced the Simple Housing Acceptable Solution (SH/AS1), as part of the NZBC. SH/AS1 made reference to expansive soils and the soil classes as per AS2870:1996, which at the time was the current version. This reference to expansive soils and soil classes, also applied scaling factors from SR120A to the upper limit ys values of each soil class to account for a 500-year return period, similar (but slightly varied) to the recommendations of SR120A. To my knowledge, these scaled ys values were not (prior to 2019) referenced in any other NZBC documentation except SH/AS1, and I had never seen these referenced in any geotechnical assessment of expansive soils, nor any structural foundation design.
- Shrink-swell index (ISS) ranges corresponding to each soil class (as noted in SR120A) were also referenced in SH/AS1, although these do not appear to be scaled in any way. SH/AS1 also had prescribed solutions for concrete slab foundations on expansive soils, however the scope of SH/AS1 was limited (to simple housing), and the prescribed slab foundations did not allow for typical raft/waffle type slabs that are common within the New Zealand industry.
- In 2011, AS2870:2011 was published in Australia. Important changes that affected the day-to-day design of raft slabs were: the split of soil Class H into Class H1 and Class H2; and an adjustment of some soil class ys value ranges. AS2870:2011 outlined soil classes and ys values as Class S (<20mm), Class M (20-40mm), Class H1 (40-60mm), Class H2 (60-75mm), and Class E (>75mm).
- As AS2870 was a referenced document in NZS3604 and the NZBC at the time, it stands to reason that this new version effectively made the soil class references in SH/AS1 out of date. Certainly, geotechnical and structural engineers in NZ began referencing the updated AS2870, especially the new soil classes and corresponding ys values (with no reference to increased ys values by way of scaling factor). SH/AS1 was never updated to align with AS2870:2011.
- In 2018, MBIE decided to revoke the Simple Housing Acceptable Solution due to its low uptake. However, MBIE noted that some of the content was applicable to other residential buildings; specifically, the reference to soil classes and the corresponding scaled ys values and ISS values. Therefore, in the November 2019 NZBC update, this information was transferred to B1/AS1. Unfortunately, this was simply a transfer of existing information, and it appears to have been made with no consideration for the changes to AS2870 in 2011. I believe that this was an oversight, and that there was no specific intention to differ from the soil classes in AS2870:2011. I also note that the references to AS2870 within B1/AS1 specifically refer to AS2870:2011. This transfer of information also included the prescribed slab foundations from SH/AS1, but these solutions did not allow for typical raft/waffle slab construction and are limited in their use.
Unfortunately, it appears that many structural and geotechnical engineers have not understood the error regarding soil class references, and it has caused much confusion since the 2019 NZBC update. One reason that this error is such an issue is because it has effectively rendered large parts of AS2870 obsolete. AS2870 has prescribed methods for foundation designs, with much of the information tabulated based on soil class; these now do not align with the soil classes within the NZBC.
For example, if a geotechnical engineer was to provide expansive soil information as per the current NZBC, they may note that the soil class for a specific site is Class H with an upper limit characteristic surface movement of 78mm. The shrink-swell testing for the site (ignoring the issues that the T+T paper raises), may actually have provided a calculated site ys value of say 48mm, which is slightly higher than the Class M limit (but most often, we see geotechnical reports refer to a soil class and its corresponding upper limit, rather than the actual calculated site ys value). As Class H does not exist in AS2870:2011, and since 78mm is greater than the upper limit of Class H2, in this case if a structural engineer was to use a prescribed foundation solution from AS2870:2011, they must use a solution for Class E soils, despite the calculated site ys value being toward the lower bound of Class H1. Of course, this would give a foundation solution that works, but it would be at a significant unnecessary cost.
Purely based on this logic, I would suggest that B1/AS1 be updated to reflect the soil classes of AS2870:2011. However, there is another concern that I believe would also need to be addressed, specifically in regards to the scaling factors that have been applied to the corresponding ys values of each soil class within B1/AS1.
BRANZ SR120A recommends that a 300-year return period be adopted for serviceability limit state (SLS) design of foundation on expansive soils (with a scaling factor of 1.0 applied to ys values), and a 1000-year return period for ultimate limit state (ULS) design (with a scaling factor of 1.2). The information on expansive soils that has been added to B1/AS1 (November 2019), explicitly states that a 500-year return period for SLS design has been adopted (with a scaling factor of 1.11), deliberately differing from the SR120A recommendation of 300-year return period. As this is a clearly noted decision, this point is not debated.
I believe that the scaling factors have been applied within B1/AS1 incorrectly, and that this was incorrect from the outset of the Simple Housing Acceptable Solution in 2010. This is based on the following logic:
- SR120A states that a 300-year return period would require a scaling factor of 1.0 (i.e. no change to the soil class ys values of AS2870).
- Considering standard probability/risk and return period philosophies, it would be expected that a 500-year return period event would be greater (worse) than a 300-year return period event. Therefore, if we were to consider a 500-year return period for design, our foundation solution would be expected to be more robust (stronger/stiffer) than if we were designing for a 300-year return period.
- SR120A repeatedly states that scaling factors should be applied to the calculated (site) ys values (there is no reference to scaling of soil class limits/ranges); i.e. it is implied that if a geotechnical engineer was to carry out a shrink-swell test for a specific site, and calculate ys values by the methods noted in AS2870, then it is this value that should be multiplied by the scaling factor. It would be expected that this scaled site ys value would be compared to the unaltered ranges of ys values for each soil class, and then a soil class determined accordingly.
- For example, if a geotechnical engineer calculated a (unscaled) site ys value for a specific site of 38mm using the methods noted in AS2870:2011, and compared this to the soil class ranges of AS2870:2011, a soil class of Class M would be determined (based on Class M ys values = 20-40mm, as per AS2870:2011). However, considering the conversion to a 500-year return period by using a scaling factor of 1.11, the scaled site ys value becomes 42mm; this would result in a Class H1 soil being determined (ys values = 40-60mm). In this case if a structural engineer was to use a prescribed foundation solution from AS2870:2011, a Class H1 solution would be adopted which would be more robust than the Class M foundation solution that would have otherwise been adopted if no scaling factor was applied to the calculated site ys value.
- From what I have seen, I believe that many geotechnical engineers are not applying scaling factors to their calculated site ys values. This may be because they consider that the scaling factors have been applied correctly to the soil class ranges in B1/AS1. Running through another example similar to above, but with the B1/AS1 values considered, if a geotechnical engineer calculated a (unscaled) ys value for a specific site of 42mm using the methods noted in AS2870, and compared this to the soil class ranges noted in B1/AS1, a soil class of Class M would be determined (based on maximum Class M ys value = 44mm, in B1/AS1). If this calculated site ys value was compared to the ys values in AS2870:2011, a soil class of Class H1 would be determined (ys values = 40-60mm). The intended effect of increasing from a 300-year return period to a 500-year return period and increasing robustness, has in fact resulted in a less robust solution.
- Based on the above, I consider that the scaling factor utilised within B1/AS1 has effectively been applied to the “wrong side of the equation”, and is having the opposite effect than intended. As a result, I believe that we may effectively be designing for a return period closer to 100 years (1.00/1.11 = 0.90, which is close to the scaling factor for 100-year return period = 0.88, as per BRANZ SR120A). This error erodes any redundancy that may exist in foundation designs, and that may otherwise help to counteract the flaws of the shrink-swell testing noted by the T+T paper.
- I am aware that some geotechnical engineers are applying scaling factors to their calculated site ys values. However, these values are currently typically compared to the soil classes outlined in B1/AS1, therefore the scaling factors are essentially applied to “both sides of the equation” and therefore cancel each other out. The net effect is that the solution is no different than if scaling factors were never introduced (i.e. a 300-year return period is effectively adopted).
Recent discussions that I have had about these concerns with geotechnical engineers have been variable. Most have generally understood the concerns that I have raised to some degree, but many have not been willing to provide information for foundation design purposes outside of what is now being dictated by the NZBC. This is causing issues for structural engineers when attempting to design foundations as per AS2870:2011, as in some cases it is now technically not applicable, or can result in under-designed or over-designed foundations.
Based on what has been outlined herein, the following solution is proposed:
- B1/AS1 of the NZBC be updated by MBIE, to refer to the soil classes outlined by AS2870:2011, with no scaling factors applied to the corresponding characteristic surface movements.
- When assessing expansive soils at a specific site, geotechnical engineers are to apply a scaling factor to the calculated characteristic surface movement before determining a soil class for the site.
In the interim, before this information may be updated in the NZBC, I recommend that geotechnical engineers provide soil class information in two forms:
- one set of information for characteristic surface movement values and soil classes to reference B1/AS1, in order to align with the NZBC (albeit that the NZBC information has flaws);
- and another set of alternative information, for scaled site characteristic surface movement values and corresponding soil classes in reference to AS2870:2011 (which itself is a referenced document within the B1/AS1).
This will allow structural engineers to select how they specify foundation solutions; either to align with B1/AS1, or to AS2870:2011. This will alleviate the confusion and discrepancy that currently exists, and will provide solutions that are more aligned with the intent of B1/AS1, AS2870:2011, and BRANZ SR120A.
To conclude, I believe that the confusion surrounding the NZBC November 2019 updates (to B1/AS1) regarding expansive soils has been caused due a transfer of previous information without consideration of the updated version of AS2870:2011. The situation is worsened due to scaling factors originally being applied to the characteristic surface movement limits of each soil class in error, when they should only be applied to the calculated site characteristic surface movements.
I believe that if the geotechnical community can understand and agree with these issues, then change to the NZBC can be easier. At least in the interim, if the industry is on the same page, then it can be justifiable to provide additional design parameters that are more in line with the intent, which will allow foundation designers to adopt methods that are suitable for the New Zealand industry.
Yours sincerely
Mitchell Rooney
(Senior Structural Engineer at Markplan Consulting)
BE (Civil), CPEng, CMEngNZ
027 700 2544
(09) 415 9793 ext. 205
mitchell@markplan.co.nz