Walls, pipes, beams, trench walls - these can all affect a gage's results.  You can reduce (calibrate for?) the effect with newer gages by taking a standard count (on your gage's block) in the location where you will perform the test just before you take your shot.  You can still have problems, but It's the best you can do with a nuke gage.

dirtsqueezer...the problem is that you don't know if you have a problem or not!  When using a nuclear density gage in a trench, next to a wall, or some other obstruction, the number of return particles counted by the internal counter can vary.

The way a nuclear gage works (for density) is that radiation is emitted at the end of the source tube.  The radiation travels in all directions from the source.  The straight line between the end of the source tube and the counter is the line in which the density of the soil is determined. Radiation emitted in directions other than this straight line are not counted and do not influence the density determination, EXCEPT....
When the omnidirectional particles bounce off some obstruction or higher density material adjacent to the gage and pass back through the counter (backscatter), they would be counted as particles returned through the straight line, thus influencing the density.  If there are enough of these, your density will be off.

It doesn't always happen, but it is enough of an issue to avoid it as you don't have a convenient means to "correlate" it out for single or several density test events.

 

Thanks for you input, guys.

Do you have any actual incidents that illustrate your point?

It is important to understand the geometry of the Nuclear MC Gage construction and the type of measurement being taken.
When taking 'direct transmission' readings (the density source is placed in a 4", 6" or deeper hole), The density counts are not greatly affected by the presence of nearby trench sides, wall or pipe, UNLESS THE MACHINE IS ESSENTIALLY TOUCHING THE MASS OR IF THE MASS EXTENDS BENEATH THE TEST SURFACE, ADJACENT TO THE 'SOURCE'.   If surface test (backscatter) are undertaken, nearby masses can be significant.

What is ALMOST ALWAYS affected by nearby masses are the moisture readings. Remember that the moisture source and the detectors are in the MC gage case and do not measure the moisture at depth but near the test surface. A very moist trench side adjacent to the  Nuclear MC Gage can increase the moisture reading by several pcf.  In-trench calibration/correction is required.

Thanks for your input.  
I believe I have a firm theoretical grasp of the instrument.  I have been testing on an FHWA project which requires moisture and density counts not only for the standard count, but for each individial test.  I have observed that as the compaction increaces, these counts decrease.  
One could draw the conclusion that the denser the fill, the less radiation is able to penetrate, and the higher your pcf.  I will disregard the concept of false high readings away for the time being, because I believe those tend to be attributed to adverse chemical content of the fill- say if your soil contained H, as is the case with bottom ash from a coal plant.  
Practically speaking, then, the only way I can see to get a false low reading is if the radiation was REFLECTED by a nearby structure, such as a pipe or concrete wall.  Theoretically, this would add radiation sensed by the gauge that would otherwise be blocked by increaced density of the fill, and register as low compaction.  
Though I respect and appreciate your opinions, in the field, I simply have not found this to be the case.  A contractor will say, "Well, we're not getting compaction because of the wall".  
I will say, "Well, why don't you try a second pass?"
The compaction will inevitably increace.  I will leave with a passing compaction report, and the contractor, though he is probably counting the man hour that he lost, will probably appreciate the nice, flat road five years from now when he drives over it.  
In short, I have simply seen too many inverted speed bumps on the road to give a contractor more reason to bend the rules.  
Thanks for your comments, guys.  I appreciate it.  

Way back when I used to do field density testing we used to calibrate both inside and outside of the trench then apply a correction. I can't remember what the correction was. We would also take moisture samples back to the lab to check against the meter mc. It was the moisture that was affected. For many sites we found that the meter would read a couple or more percent high and would apply a mc correction which would of course bring the density up. Some sites have serpentine or other minerology that affects the meter. It sounds like you have it under control. Don't take any stuff from the contractors. They know all the problems associated with field density testing. You have probably been told that your compaction curve is high too! I would just politely ask them to recompact before we get into a research project and the density would amazingly come up. Keep up the good work!

Dirtdoc1

dirtsqueezer (Geotechnical) Dec 20, 2002 wrote, “Practically speaking, then, the only way I can see to get a false low reading is if the radiation was REFLECTED by a nearby structure, such as a pipe or concrete wall.”

Interestingly enough, density readings from direct transmission tests of 6” of depth or greater are NOT affected by nearby large vertical objects because the source rod is too deep in the ground to reflected photons to the detectors from nearby vertical objects.  Moisture readings will still be affected no matter what depth the source rod is at because the source for measuring moisture is in the gauge, not at the end of the source rod.  However, if a contractor doesn't like your test for whatever reason, just remind him/her that you control the pencil regardless of what the gauge reads.

From a Troxler operations manual:  “Vertical structures can scatter [reflect] neutrons and gamma photons back to the gauge, increasing the possibility of moisture or density errors due to high counts.”

In general, the lower the number of photons that reach the detectors and are subsequently counted, the higher the material’s density.  When vertical structures, trench walls, curbs, the technician’s truck, are present then additional photons are scattered or reflected back to the gauge and counted resulting in a higher count and thus a lower density reading.  This is especially true when taking backscatter tests as ALL photons counted are reflected to the detectors.

The same holds true for moisture readings.  The detectors only count thermalized neutrons (neutrons having been slowed by collisions with hydrogen).  The higher the moisture count, the higher the moisture content.  When vertical trench walls are present, additional thermalized neutrons are scattered or reflected back to the gauge resulting is a higher count and thus a higher moisture reading.

Determining offsets for trench readings either for density or moisture is a very simple process but requires reading the operations manual.

Any test method is only as good as the tech making the test and the other observations he/she has made.  Generally tests only confirm something I already know.

SCET - Techmaximus

As techmaximus pointed out the mosture reading is essentially a backscatter mode. The more moisture, the less penetration. A prctical consideration in a trench is that the soil is more damp. also as yuo compact, excess water can bleed up. Usuallyy preparing the bottom well and checkig in the center (away from the edge) should give you good results.

The moisture content of material is determined by the source within the gauge emitting neutrons which are scattered or reflected back to the detector within the gauge.  Neutrons penetrating the material are slowed (thermalized) due to their reaction with hydrogen (or other material) within the material.  The detector within the gauge is not sensitive to the fast neutrons, only the slowed neutrons.  This is why the higher the moisture content in the material being tested, the shallower the depth of measurement; the higher the moisture content in the material means the less deep the neutrons have to travel to get thermalized.

However, I disagree that trenches contain “soil that is more damp.”  I’ll give you that as material is compacted in a trench with vibratory equipment (whacker packers and the like) as is most common with trench backfill (especially narrow trenches) moisture can bleed up or rise to the top.  As vibration from the compaction equipment beats the moisture up this may make it appear that the trenches are damper and especially so in materials with higher clay contents.  I mostly use a Troxler 3430 which is super easy to set trench offsets with.  Not setting offsets or determining correction factors for trench work simply leads to erroneous test results.

SCET - Techmaximus

Trenches for a variety of reasons, weeping clay, limited drainage paths and just generally damper conditions can have higher moistures at the surface. My point was and, techmaximus, I concur with you that the Troxler can be used sucessfully in trenches with proper technique and common sense, but good results do require a little attention to details.

Personally, I like to confirm from time to time and especially in important situations the actual moisture by taking a sample to the lab. I did an extensive comparison of our troxlers and lab moisture contents (years and years ago) and found that the lab sample was dryer by about 20% or so - if my grey cells are still working).

BigH is right and points to another issue with this.  

Techmaximus...you are assuming the only error in density has to do with the moisture reading of the nuclear density gage.  This issue was around long before nuclear density gages had built-in moisture capability.

The primary issue is one on unpredictability of the result, due to an unpredictable response count from the device.  It counts particles that make it back to the counter within a time domain.  In an idealized condition, the particles are transmitted directly through the soil to the counter, thus giving an assessment of density.  If the particles are scattered by obstructions or concentrated by obstructions, then erroneous results occur.  In one case conservative, in one case not.

To BigH's point...correlations should be done, both for moisture and for density.   Further, if the sitework exercise is large enough, there should be a site-specific correlation done to give a greater level of confidence of the results.

To make this topic more interesting, please read the US army Nuclear Gauge manual for MC-1 on page 1-20:

It says: the good practice in a confined space would be face the tester logitudially in the direction of the trench and across the trench for moisture determination.

Does this make sense? It looks like to me in the field we always place the NG logitudially in the direction of the trench, and read the Soil Moisture and density at the same time. We never tried or had the idea to get the SM in another direction.

US army Nuclear Gauge manual for MC-1:

(http://www.dod.state.hi.us/HIARNG/safety/documents/radiation_safety/TM%205-6635-386-12&P(MC-1).pdf#search='CPN%20MC3%20Gauge%20operating%20manual')