LRFD Design for Geotechnical Projects 1

[bushel3]

As recently discussed in the Sheet Piles Retaining Wall LRFD Thread, the DOTs will require the use of LRFD design by 2007 for geotechnical structures. I have not yet designed anything using the LRFD criteria. I have heard that the LRFD method will give a more conservative design when designing caissons/drilled piers or piles because the phi factors in AASHTO are too low.

In your experience will you come up with a similar design using LRFD compared to the allowable strength design methods currently in use? If not which one is more conservative?

 

[BigH]

From what I keep hearing, the factors that will be "used" in the LRFD designs will be "tempered" to give, generally, the same answers as in our normal designs. Maybe not always, but that is what many of the proponents are saying.

 

[jdonville]

bushel3,

If I understand the LRFD concept correctly, the degree of conservatism with respect to ASD approach will depend on the importance of the structure and the method(s) used to determine the design load and available resistance. See the AASHTO and FHWA LRFD sites for more information.

Jeff


Jeffrey T. Donville, PE
TTL Associates, Inc.

http://www.ttlassoc.com

The views or opinions expressed by me are my own and do not necessarily reflect the views or opinions of my employer.

 

[bushel3]

jdonville, you are correct that the degree of conservatism depends on the type of structure, the type of soil, the method of testing the soil. Our compnay did 1 job last year where we designed piles for a bridge using the "old" way and using the LRFD in AASHTO. The design using AASHTO was considerably more conservative.

The place where my boss in particular feels that AASHTO is too conservative is on the "performance factors". The performance factors are between .35 to .9 with most of the values being between .5 and .7. AASHTO table 4.10.6-1&2 show the performance factors.

I'm not so sure I agree with him because when you divide the performance factors by one 1/.35 = 2.86, 1/.5 = 2, 1/.7 = 1.42 and 1/.9 = 1.11. When you multiply these results by the load factor which is usually 1.3 (AASHTO 3.22) to get the actual factor of saftey used you get (respectively) FS = 3.71, 2.6, 1.85 and 1.44. These factors of saftey appear reasonable to me.

My boss argues that there shouldn't be any performance factors as low as 0.35 and in my opinion gets hung up on those low performance factors. He plans to continue using the old methods until the performance factors, particularly the low ones are increased.

Anyways I was looking to see if anyone else has compared the two methods and noticed a trend in which was more conservative or if they typically give similar results.

 

[Panars]

First, the 2006 interims just came out for the AASHTO LRFD specifications, and Section 10 for Foundations was replaced in its entirety. My comment is based on these interim revisions.

The advantage to LRFD that I see (at least for deep foundations) is that the resistance factors are related to the amount and type of load testing that is performed. For example, if you design a pile foundation based soley on SPT blow counts, the resistance factor is 0.3. However, if you perform dynamic testing of the piles with a PDA and signal matching (similar to what Ohio DOT performs), the resistance factor is 0.65. Assuming a load factor of 1.3, these factors roughly correspond to safety factors of 4.3 and 2.0. So, if you are not performing any load testing, then yes the LRFD design will be more conservative.

I think the relating the resistance factors to the load testing is a good thing, because it will allow the geotechnical designer to quantify the benefit gained by load testing and justify it in terms of dollars to the client.

 

[bushel3]

Panars,

That is a good point that the LRFD method will allow you to justify more testing to your client. Currently it is hard to quantify how the additional testing will improve the design and save cost.

However, is it necessary to for AASHTO to quantify what factor of saftey we should use. I think the freedom to make an educated choice in the factor of saftey based on total knowledge of a particular project is great. It frustrates me that AASHTO has to dictate the FS based on an incomplete picture of each project, (granted the quality of the testing is a major factor into determining my FS's now.)

I was taught that the factor of saftey should correlate to the degree of confidence that you have in your answer and the price you'd pay if the structure failed. The method of testing (spt, cpt, etc.) plays a factor into that but that is not all that plays a factor my choice of a factor of saftey. I also consider what will happen if the structure fails (very important or not), how many similar designs on similar material I have sucessfully constructed before, how many borings we did (regardless of quality of testing afterwards), and so on. Generally I pick a FS between 2 and 3, normally closer to 3. When we load test pile we design for an FS of 2, but only load test to an FS of 1.5.
I rarely design anything with a factor of saftey over 3 unless I have almost no data to go on or I'm not confident that all the parameters of the project have been fully laid out.

 

[jdonville]

bushel3,

My understanding of the situation is that the resistance factors are based on a statistical analysis of both the reliability and repeatability of the particular test method (e.g., SPT, CPT) employed during the investigation. AASHTO has made a commitment to update these factors as more research indicates that changes should be made.

It is important to remember that the LRFD approach explicitly tries to better quantify the unknowables associated with particular investigational tests and computational methods in order to keep the design methodology relatively simple. Otherwise, designers might need to perform Monte Carlo simulations to justify their designs to the agency funding the project.

Naturally, increased quality control monitoring, such as actual production load testing of individual founation elements, translates into more feedback into the design and increased confidence with the ability of piles, to use Panars' example, to provide sufficient resistance to the design loads.

The reason AASHTO gets to decide which FS to use, is that the public agency has a twofold agenda (at least):
1) The structure should perform satisfactorily over the design life (increases conservatism) i.e., should be usable for decades if properly constructed
2) The structure should be designed to be as economical as possible (decreases conservatism) in terms of material, labor and operating costs

The FS used by the designer (or mandated by AASHTO or the lead funding agency for a public project) is at best an estimate of the true FS and is used to balance the competing objectives of points 1) and 2) above. If you don't want to use LRFD after mid-2007, don't bid on DOT projects.

Jeff


Jeffrey T. Donville, PE
TTL Associates, Inc.

http://www.ttlassoc.com

The views or opinions expressed by me are my own and do not necessarily reflect the views or opinions of my employer.

 

[rochplayer]

Not to be too nit-picky, but there are no safety factors in LRFD. There are only Load and Resistance Factors. Load factors account for uncertainty, criticality, and application of applied loads. Resistance factors vary based on instrumentation, testing, design method, data quality and quantity, etc.

The basic LRFD equation is:

sum(gamma_i * Q_i) <= phi * R_n

Where:
gamma_i are load factors
Q_i are loads
phi are resistance factors
R_n are unfactored nominal resistance


For example, the dead loads on a structure are more readily quantifiable, so the load factor for applied load of structural components where the applied load for driving (e.g. overturning) is 1.25, but a dead load considered in resisting (e.g. in sliding) is 0.9. Live Load on the other hand is 1.75 for driving and is neglected in resisting.


LRFD also uses multiple loading combinations which are evaluated to determine the most critical case. The three general loading categories are Service, Strength, and Extreme.

The Service Limit case relates to general use, or service, of the particular structure. Service loading cases are generally limited by deformations that would cause the structure to fail in its intended use, but not fail structurally. Service limits can include settlement (e.g. maximum settlement of 1"), overturning, lateral displacements, global stability, etc.

The Strength Limit state is used to ensure that a structure will not fail under significant loadings that are expected to occur during its design life. It may settle more than desired, for example, but ultimate bearing isn't exceeded.

The Extreme Limit state is used to make sure the structure survives under an extreme event such as a seismic event, extreme scour, vehicle or ship impact, etc.

Each limit case has multiple load combinations which are evaulated and compared to the factored resistance of the foundation.

Resistance factors vary with limit case. For example, with drilled shafts designed in Washington, per the WSDOT, the resistance factors are as follows:

Service limit (capacities per Reese and O'Neill with settlement limited to 1-inch)- side friction = 1.0, end bearing = 1.0
Strength limit - side friction = 0.55, end bearing = 0.50
Extreme limit - side friction = 1.0, end bearing = 1.0

Resistance factors for Strength Limit in AASHTO for side resistance of drilled shafts can vary from 0.65 in clay to "using judgement and available experience" in sands and gravels. With a Load Test, AASHTO allows resistance factor of 0.8. Higher resistance factor because of more confidence in the nominal capacity.

So LRFD does not equal ASD. It has a different terminology and way of looking at soil capacity. The basic analysis of soil is the same in both method, but how the results are applied is different.

 

[Panars]

rochplayer-

You are correct, and that was a very good summary of the LRFD method. However, when trying to explain a new concept to someone, it is often beneficial to put the new concept into terms the person is already familiar with, even if the correlation is not strictly correct.

To help geotechs understand LRFD, I will often compare the combination of load factors and resistance factors to the safety factor they are already familiar with. Of course, then I have to explain the differences and why the comparison is not as simple as it may first appear.

 

[bushel3]

rochplayer/panars

Numerically, there is no difference between using a factor of saftey or LRFD. Both methods have the same result, but come from different theories. Both methods reduce your design strength resulting in larger members/footing, etc. Both methods result in the design being safer than required. I don't care too if one method is more theoretically correct if the end result isn't improved. What I was originally looking for when I started this post is reasons why we should change methods to LRFD in terms of engineering economy.

I was taught LRFD for designing steel and factor of saftey for solving geotech problems. I can do either method and will do whatever one my client requests. I might be too practical for my own good, but why change methods when both provide similar results. It appears to me that the new method (LRFD) is even more conservative than the old in many cases. LRFD seems to define the "factor of saftey" or the statics/reliablity in a more refined way but in the process it complicates things a little more.

I was told that the reason LRFD is used instead of ASD in steel design is that LRFD results in smaller members or more economical members than ASD. That is not a bad reason. So far in my limited experience with LRFD in foundation design it seems like LRFD does not result in a more economical design. Not to mention, it adds another step/varioble to each problem. Additional steps result in additional opportunities for mistakes.

The other thing that I like about using a factor of saftey in foundation design is that makes it very easy to see your mistakes. I can divide an ultimate strength by 2 or 3 in my head to do my reality check on my designs. It is much more difficult to multiply one side of the equation by 0.45 and the other by 1.3 while doing a reality check.

I suppose the point that I'm driving at is our economy rewards speed of design. It also rewards the best design, but you can not get a job if your price for design is twice your competition's price. It does not matter if you will save 25% of the overall project cost because most will not pay for your expensive design.

I have a basic (not great) understanding of many areas of engineering. To me it seems like most of our modern improvements to our methods take longer to do. They are more theoretically correct, but add many new variables and steps to design. The new concrete code seems to have 10-15 new such refinements. Each one adding at least one more step. This is fine if you can bid a job on the final cost of building your design and not the time for design.

The other problem with additional time for design is that on smaller jobs the additional time will not result in any or much savings. These refinements only help when the project is large and using 1 less bolt per connection times 10,000 connections = $100,000 savings. Losing the one bolt may make it cheaper but it also may make it less durable. Needless to say I'm not a big fan of the frequent minor changes I see happening and I currently have lumped LRFD into this category.

LRFD does not appear to me to provide enough improvements over using a factor of saftey to justify the additional variables. My unresearched thoughts. I'd enjoy hearing your responses.

 

[rochplayer]

I think one of the issues I have with LRFD/ASD and geotechnics in general is the issue of conservatism heaped upon conservatism. Do not misunderstand - conservatism is wonderful, but a problem with moving to LRFD from ASD can be best illustrated as follows:

Suppose I am designing a bridge. I perform a site investigation using classical techniques - Mud rotary drilling and sampling with SPT. I drive some LPT to collect additional sample for mechanical testing. Suppose the bearing layer for a deep foundation is a SM - Silty Sand with Gravel. Average N'60 for the layer is 30 blows. Correlations to SPT give a range of phi = 34 - 40. Because of the gravel in the sample, the blow counts may be overstated. Hmmm, 40 seems pretty high for a N'60 = 30 soil. So I choose a design phi = 36 to be conservative.

In my opinion, the standard ASD safety factors were developed with the above thought process. Now, if I plug this value into ASD, I come up with an ultimate capacity in skin friction and bearing and apply a SF = 2.5 because I will be doing dynamic testing with a PDA. Settlement review shows within structural tolerances.

For LRFD, the resistance factors are supposed to take the uncertainty of the site investigation methods into account. Design based on SPT has one resistance factor, CPT another, Wave equation another, and load test results another. The resistance factors also vary based on analytical method and the uncertainties associated with the methods. The question to me is then, which soil parameter to use? Which end of the spread for phi do I use in the SPT correlation if I am going to multiply the resistance by 0.45 per AASHTO for skin and end bearing of piles, plus multiplying the load by a load factor? I am still planning on using PDA (lambda = 1.0), but I end up with an "equivalent" SF = 3 (1/0.45 * 1.35). I will have a more conservative design if still use the lower bound value of phi=36 based on SPT.

For effective implementation of LRFD to geotechnical engineering, I believe it is going to take a change of thinking in how we address soil properties and uncertainties. The touted power of LRFD is that the uncertainties can be broken down into their individual components and be evaluated more rationally that a gross SF applied to the end result. But, if we still don't know all that much about the soil's behavior, does LRFD provide much value?

On the LRFD/ASD issue I am still on the fence. I can see the power of LRFD, but in my opinion for it to be used to its full potential, it will require more up-front cost (additional testing and investigation) to be able to firm up what the soil conditions "really" are (as well as be told by a 6-inch borehole logged by an E1).

 

[BigH]

We have had some very good discussions on the LFRD topic in past threads - do a search for them. Focht3 is dead set against; VAD is for. I am inclinded to be against especially since proponents have actually come out and said they 'temper' their factors to be in line with the traditional geotechnical methods. I'll look up some articles I have and report later.
Ciao - and Have a Great Year of the Dog!

 

[BigH]

Try the following: Maybe more to come. Washington GeoManual is good (Chapter 8 as noted).

http://bridges.transportation.org/sites/bridges/docs/AASHTO LRFD Geotech Roadmap - JAD Version.doc

http://www.washto-x.org/file_search/file_display.php?course=31

Gordon A. Fenton, D.V. Griffiths, and W. Cavers
Can. Geotech. J./Rev. can. geotech. 42(5): 1422-1436 (2005)

http://www.engmath.dal.ca/tc32/OsakaReport2005.html

http://www.wsdot.wa.gov/fasc/EngineeringPublications/Manuals/2005GDM/

http://www.wsdot.wa.gov/fasc/EngineeringPublications/Manuals/2005GDM/Chapter8.pdf

(Specifically)

 

[BigH]

http://trb.org/publications/nchrp/nchrp_rpt_507.pdf

Deep Foundations
168.166.124.22/RDT/reports/Ri03030/or06010.pdf

http://www.eng.nus.edu.sg/civil/people/cvepkk/CPaper_IWSKamakura_2002a.pdf

by Kulhawy and Phoon Good Background

http://www.eng.nus.edu.sg/civil/people/cvepkk/Special_KGS_2004.pdf

by Phoon Good Background

http://www.modot.org/business/manuals/documents/3.80_LRFDtitle.pdf

http://sections.asce.org/indiana/2406TechnicalSummary.pdf

http://www.jcss.ethz.ch/Events/WS_2002-03/

Paper_PPT_WS02/HONJO_partial_factors-JCSS.pdf

http://international.fhwa.dot.gov/Pdfs/Geotech.pdf

 

[VAD]

bushel3 and others:

As BigH has said, this topic has seen previous debates.Some of us are for and some against. Indeed many of the perforfance factors result from the FOS. As Geoff Meyerhof once said we have not made any significant strides in fine tuning these factors and therefore the Geotechnical Engineer would still be providing values that we have been accustomed to provide although they may be in a different format. Others have said that the Factor of Safety is a factor of ignorance, which is true as well to some extent.

We were brought up to use these factors without a true understanding of how they were derived except to know that from past experience those before us used these with success. We were not fully trained in how to decide on what numbers or values to pick for parameters. Some of us would take the lowest value, others the mean, some of us if we were not confortable with the test results would disregard the numbers and use some value from experience or from a text book.

If a great number of values could be had then we would engage in statistical evaluation of different types and orders. In the end we may even temper the values with judgement. As a few engineers to choose values for strength or fricyion angle from a set of test data and see how varying the numbers would be.

In the past we have provided values to structural engineers and they have used same to size their foundations, and decide on how many foundation units are required to support a structure of given loading. When the structure was completed we stood up we said we did a good job.

Many of us were perhaps not aware of what the applied design loadings were, whether there was considerably lighter loads etc and probably the foundation unit was the most practical irrespective. This feeling of having done a good job I think is early in one's career. How often have we not found out later on the the structural designer also lowered our numbers for his peace of mind.

LRFD in my opinion brings geotechs and structural engineers to communicate with each other which should be the ultimate goal of any bridge or building infrastructure design. Too often geotechs work in providing a report and not intimately involved in seeing the project through, which is rather unfortunate.

The Code that governs structural work oulines what load factors are to be used but do not necessarily say that you cannot modify these. The choice of 1.25 for dead load is not fixed. For some situation 1.00 can be used and even 0.9. There are provisions for choices. One has to remember that some of the numbers apply to certain loading conditions that structures are exposed to and the designer has to state his case for use of these factors.

The factors used for a bridge on the Interstate would not be the same for a bridge on a secondary highway. One has to be careful if the bridge on the secondary roadway will some day be subject to higher loads and warrant higher load factors.

At least there is on the structural side some approach which is laid out and can be reviewed, argued about and defended etc. On the geotech side, this should be the objective as well. When we temper results that we have obtained to decide on a value and have applied judgement to it, this needs to be stated as well. In so doing we can encourage dialogue between the geotech and structural engineer. This will enable to see what the structural Engineer and the architect face when loads that have to be applied to certain grounds have to result in more expensive structural schemes. How may times have you not been asked if you can see if the soils can have a higher bearing capacity and how many times have you not teased the values a bit higher. Another FOS.

I do not necessarily agree that the code factors whether load or resistance are the end all. As I have said for geotech work they have been derived from FOS used previously. I think that we should strive to understand the soil conditions to the utmost and make the best choice of values first. There is a factor for the equipment we use in the soil exploration, the tech on the drilling rig, the lab that does the testing etc. How many times have we not agonized on results. Well it is time we state that out in our reports and in our premise for our choice of values.

The LRFD sets the stage for meaningful discussion between the geotech and structural engineer and at least it allows some of the blackbox from both sides to be out on the table. At least both sides can walk away with a better understanding, appreciate what risks to take and what not to take in a collective way.

Further this stage allows for others in future years to see the premise of how we have decided to choose our parameters and can decide that when a structure needs to be upgraded whether the values given can be relied upon or whether there is even a margin to use some higher value and have an overall influence on costs etc. How often have not been at a loss to know whay certain values were chosen when we are asked to evaluate the load carrying capacity of an existing foundation. How often do we not disagree. How many of us are confident to say the foundation can take additional load when we are unable to undertake a load test.

What really puzzles me is that despite that we do the most expensive load tests on piles and get the most beautiful load displacement curves we still resort to applying a FOS of 2. Why?. Is it because we are still uncertain and this is what we have been trained to do. Certainly I should be able under extreme conditions to invoke a higher load if I understand my site conditions. This to me is where we should be going. This is the challenge of geotechnical engineering. Loads must be used in conjunction with soil derived strength etc. the beauty is the iteraction between load and soil and the ultimate behaviour taking all extraneous factors into consideration. The geotech and structures specialists have to be in sync.

As I said in other threads, the LRFD is here to stay and it is time to get on board for a real adventure in the true understanding of the behaviour of the ground in relation to applied loads.

I have said enough. Excuse the length

[Cheers]

 

[Panars]

Thank you VAD. I feel much the same as you. If nothing else, LRFD will get the geotechs and structural engineers to communicate more. This is a good thing.

At the risk of being called a nitpicker, I will clarify one minor thing you said. "The factors used for a bridge on the Interstate would not be the same for a bridge on a secondary highway." For AASHTO in the USA, the loading might be different for a bridge on an interstate compared to a bridge on a secondary road, but the load factors would be the same. The only difference in factors might be the Operational Importance Factor, which can vary from 1.05 to 0.95. However, many states are considering using a value of 1.0 for all bridges and effectively ignoring this factor. (As I said, a very minor issue)

 

[DRC1]

It has seemed to me that since structural steel engineers moved away from ASD to LRFD (concrete has been there for years)that someone felt soils should be engineered by LRFD for some sort of compatibility. I disagree.
ASD sweeps all the loads into one FOS. LRFD recognizes that given a high certianty regarding loads and transfer mechinism, load factors could more economically used to reliabllly design strucutes. For steel and concrete, material properties, load tranfer and applied loads can be well defined and LRFD is certianly a design option.
For soils, we still struggle to understand load transfer, materials are not homogeneus, and testing is often emperical, and how loads are appled is not as clear cut as in framing loads. Because no matter how well we try to understand the soil we work in, much of the design work we do is based on emperical methods. Further, to utilize LRFD, more soil testing must be done than is typically done now (not that would be a bad thing no matter which method is used)
ASD and LRFD should be considered seperate tolls availble to the designer. A carpenter has many tools because none will do every job. I feel that LRFD should not be mandated because more often than not, ASD would be a better choice. Yet if LRFD is available and used, it like ASD will evolve and become more useful. I say make it option, but not the only method.
It is interesting that AISC now is coming out with a combined spec for ASD/LRFD