As I continued reading the sand chapter in Bryan Higgins’ Experiments and Observations Made with the View of Improving the Art of Composing and Applying Calcareous Cements (1780), I came across this paragraph:
“When sand was poured into the glass cylinder until it was filled , and water was added before the sand was packed, by a slight agitation of the vessel the sand contracted in a much greater degree than is above expressed. Upon the whole it seemed that water, by poising the grains, facilitates their sliding on each other to fit well and fill the spaces.”
The first time I heard about a change in volume in wet versus dry sand was when Morgan Phillips asked me to read an article he was preparing titled A Source of Confusion About Mortar Formulas. This article was subtitled “In order to prevent errors, more specificity is needed for proportioning and mixing mortars.”
In this 1993 article, Morgan was discussing the problems of mixing mortar ingredients from dry measure and the differences in volumes between the dry and wet product. He states in the section Bulking of Sand: Another factor that can compound the problem:
“This is the increase in bulk volume of dry sand that occurs when water is added. If sand is measured dry, more actual sand is put into a mortar than if the same volume measure of damp sand is used.” *
He goes on to conduct some experiments with a particular sand and further along in the article he quotes instructions in the ASTM C-270, section that states that “when necessary, sand quantities should be adjusted to provide for the bulking of the sand.” Following this is a table shows the formulas for sand as:
15 measures (dry) x 117% = 17.6 (measures damp) *
We were both working on the Wyck project in Philadelphia and I had just begun working with lime mortars and renders. I did not fully understand the reasoning for this problem with sand, but decided that I should switch to alcohol to measure the void space in aggregates. My understanding then was that surface tension of water might be affecting it and alcohol would give a more accurate reading. Recently I have questioned my practice of using alcohol instead of water and decided to conduct some experiments to see what I could learn. *As you’ll see below, my findings are the opposite of both Morgan and ASTM’s statements (starred above). Has a typo in the past just been repeated and magnified?
The basic idea behind determining the void space in a mortar (and thus the ratio of lime needed to adequately cement the sand) is that you take a given amount of sand, and add liquid until the sand column is saturated with no additional liquid on top.
For this particular sand the amount of alcohol needed was 34 ml meaning that 100 ml of dry sand has air space of about one third of its volume. This then equates to a lime putty-to-sand ratio of 1:3.
But does it really matter whether I use water or alcohol (200-proof ethanol)?
I repeated this test with both alcohol and water several times with the same sand the results were within 1 ml each time.
Next I filled two beakers to exactly 200 ml with sand from the same container. I added a measured amount of water to both, but stirred the sand in the beaker on the right. Then I tapped both beakers several times against the bench top.
The un-stirred sand decreased in volume from 200 ml to about 185ml.
The stirred sand decreased to 175 ml.
The difference in the packing density is evident. But even more interesting is the difference in the the pore space based on the volume of water required to wet each sand. The 185 ml sand took 68 ml of water, the sand at 175 ml needed only 55 ml. In the first case, I added water slowly until the sand was just wet, whereas in the latter case I poured in 100ml and then stirred it before pouring off the excess water, measuring what was removed.
The difference is obvious. The stirred sand has reorganized and efficiently packed the different sized particles. This causes the void space to decrease. It seems to me that we should be basing our lime to aggregate ratio on a well packed sand that represents the relationship of the particles in a cementatious environment.
It is also interesting to note how dense the compaction is when the water was stirred. As you can see, the unstirred sand at bottom will still take a fingerprint, whereas the stirred sand is so compact I could not make an imprint.
I guess if we think of any trip to the beach and the density of the sand as the tide moves out compared to that further up the shore, none of this should be a surprise.
Wet, unstirred versus Wet, stirred compaction (not out of focus)
Another thing I’ve learned from this experiment is the reason why our calculations for the amount of stucco needed for a given area often falls short. I have repeated this experiment with several different sands and have volume reductions between 12.5% and 17%. Since the sand volume decreases when wet it is important to determine shrinkage percentage for the particular sand to be used.
For instance, if you needed 100 cubic feet of mortar (as measured from dry sand), your mix, depending on the pore and void space in the sand, your finished mortar could be as little as 83 cubic feet. If you (inadvisably) used crushed rock, it could be as little as 75 cubic feet.
Looking back at the crushed rock discussed in the previous post, I decided to see what would happen when it was wetted out. The result was quite shocking. There was a full 25% reduction in volume when water was added and stirred in.
Going forward, I am going to make up some mortar samples with pore space measured both by the stirred and unstirred method and set them aside to cure. In a few months we will have some testing results from these samples. I’ll also repeat these pore and void space measurements with water at room temperature versus that between 37 and 39°F when water is densest to see what affect if any it may have.