ejge paper 2004 -0445



Hydro-Compaction of Sand Fills


Fathi M. Shaqour

Assistant Professor, Department of Applied Geology and the Environment,
University of Jordan, Amman, Jordan
e-mail: Shaqourf@yahoo.com, f.shaqour@ju.edu.jo



Compaction of backfill materials is normally carried out by mechanical means, to achieve a specified percentage of compactivity. In some cases, hydro-compaction is used for the same purpose.

The present study discusses the results of field and laboratory experiments carried out on sand fills, by application of water over the sand. Field experiments were conducted in two pipe laying trenches, with two layers of sand 1m and 0.6 m thickness respectively, while in laboratory experiments, a 200-liter drum was used. Field density tests and moisture contents were measured in the trenches after two periods of draining (2 and 18 hours). In the drum tests, the same sand was tested, but also after being washed from fines.

The results indicate that hydro-compaction failed to achieve the required percentage of compaction (95%) for structural fill. About 40% of the tests meet the requirement for non-structural fill, however the high heterogeneity of the results make it difficult to adopt the method for compacting even non-structural fills. Laboratory experiments gave slightly better results, yet not 95%. Washed samples gave even better results, although the fines content does not exceed 5%.

Keywords: Compaction, sand fill, hydro-compaction, moisture content, maximum density


Soil compaction is a process where soil particles are packed more closely by reduction of volume voids, resulting from the momentary application of loads such as rolling, tamping or vibration. This process involves the expulsion of air from the voids without the moisture content being changed significantly (Bell, 1993). It also involves the reduction of volume leading to increase of density and shear strength of the soil (Duncan, 1990). Effective compaction is normally achieved by compacting constituent particles of in-situ soil or borrowed fill to a denser configuration that can reach maximum at optimum moisture content. Well compacted materials display engineering properties that are substantially superior to the same materials in a loose state (Watson and Burnett, 1995). Soil compaction is common in civil engineering practice mainly in highway and rail road works, run ways, earth dams, sanitary landfills and backfilling of excavations for substructures and retaining walls. The quality of compacted fill is controlled by interrelated factors such as soil type, thickness of compacted layer and the compaction effort. Soil type including; grain size, gradation and clay or fine content, play an important role in soil behavior under compaction. Coarse-grained soils exist in inter-granular contact and their compaction involves particle rearrangement. The way in which these particles are arranged within the soil mass, and the distribution of particle size throughout, will ultimately determine the degree of compaction and hence the density, stability and the bearing capacity of the soil. Therefore the maximum compaction can be achieved by the best packing of well-graded soil where the fine grains fill the spaces between the large grains. However surplus of fines can be detrimental by preventing inter-granular contact between course particles.

Moisture content is a major controlling factor for soil compaction specially the fine-grained soils. Increasing the water content of soil under compaction effectively increases the workability of the soil and enables higher density to be achieved. However, further increase of the percentage of water content beyond a certain critical value will result in fall of density due to trap air and excess water in soil pores. This critical value of water content, known as the optimum moisture content can be determined by the Proctor or standard compaction test (Anon., 1976). Field tests are needed to assess the actual density achieved by compaction on site (Parsons, 1987). The interaction between density and moisture content of compacted soil is well documented in soil mechanic literature. R.R.Proctor first introduced it in early 1930s, where he developed a procedure to determine the maximum dry density and the optimum moisture content, which then became known as Proctor Density Test, and adopted by ASTM as Test method D-698 that was then modified in ASTM D-1557 (Figure 1).


Figure 1. Typical curve for the change of dry density
with moisture content (Proctor compaction test)

Effective compaction of earth materials can be achieved by applying the proper mechanical energy to a soil layer of suitably specified optimum thickness. An optimized compaction effort may be described as one that results in obtaining the compaction required from making the correct number of passes, over a placed layer of optimum thickness (Watson and Burnett, 1995). An optimum layer thickness of 200 to 300 mm is normally specified in road and sanitary line works, however the thickness of compacted layer varies with grain size and compaction effort. Generally coarse-grained soils compact more readily than fine grained ones and hence the finer the particles the less maximum thickness of layer to be compacted.

The required degree of compaction (relative compaction) is normally specified relative to the type of fill being structural or non-structural where the relative compaction is specified to be 95% and 90% respectively. The relative compaction is determined by comparing values of measured field and laboratory dry densities. Field dry density is measured by carrying out one of the standard procedures as ASTM D-1556 (sand cone test). One of the precautions against this test is the error that can be caused by slumping of the sides of the excavated density hole in case of wet granular soils and this can lead to over-estimation of the density (BS 1377). In recent years a more successful technique has been developed using radioactive instruments to determine both density and moisture content.

Backfilled materials are commonly compacted by mechanical means using hand operated vibratory compactors in case of small confined areas and machine driven vibratory compactors such as smooth and paddle foot rollers, for large open areas.

Hydro-compaction is also used by flooding or jetting the granular fill material with water in which super saturation produce degree of compaction (National Clay Pipe Institution, 1990). This method of compaction has been used for foundations with small loads. However its use is frequently held back by many shortcomings such as; risks for adjacent existing buildings, slow process, oversetting of the soils and possible cause of water table rise, and insufficient compaction to the top layers (Bagasarov et al. 1992). There are limitations and certain conditions for the success of this method such as uniformity of sand, fines content and free draining in addition to the time factor.

The present study investigates this method of hydro-compaction and discusses the limiting factors insight of results from field and laboratory tests carried out by flooding sand with water. Field experiments were conducted on sand layers of 600 and 1000 mm thickness laid in trenches of 1.0 meter width and about 30 meters length. Also laboratory experiments on smaller samples using a 200 liters size drum have been carried out.


The material that was used for backfilling of the trenches to cover the sewer lines was specified in the project to attain a relative compaction of 95% being a structural fill. Local sand was selected for backfilling and index tests were carried out for the purpose of characterizing the material. Tests were carried out, using the same natural material, which was specified for backfilling of the trenches, while washed samples of the same sand was also used for laboratory testing. Grain size analysis, specific gravity and Proctor compaction tests have been carried out. Figure 2 presents grain size distribution curves for both washed and unwashed samples, which shows very close distribution apart from the minor variation indicated by the absence of fines in the washed sample. The natural fill material is mainly composed of quartz sand particles (about 85%) and some feldspar grains (about 10%), of uniform grain size distribution, with a percentage of about 5% fine content of quartz, calcite and gypsum materials. The specific gravity of the fill material was measured for several samples and the average was 2.63 for both washed and unwashed samples. Some washed samples gave higher specific gravity as should be due to the absence of the less specific gravity fine particles. However the difference is within the error range of testing.

Figure 2. Grain size distribution of the fill material (A: Natural sand, B: Washed sand)

The maximum dry density and optimum moisture content of the material were determined by running Standard Proctor compaction test following ASTM D698. The average maximum dry density of three samples representing the natural sand and other three samples representing the washed sand was found to be 2.1 MG/m3 at optimum moisture content of 10.5%, and 1.95 MG/m3 at optimum moisture content of 10% respectively. Figure 3 presents the dry density-soil moisture curve of the fill material for both washed and unwashed samples.

Figure 3. Dry density vs. moisture content of the fill material (a: washed sand, b: Natural sand)


Two series of tests were carried out. The first series, involved field experiments on natural sand used for backfilling of excavated trenches for laying pipes of sewer lines. Two trenches with dimensions of about 30 m length, 1m width and about 1.5 m depth, have been selected for the experiments. The first one was tested after backfilled with a sand layer of 1m thickness, while the second was tested after backfilled with a sand layer of 0.6 m thickness. The second series of tests involved laboratory experiments on smaller samples of the same sand material, where 200-liter cylindrical samples (0.6 m diameter and 1m depth) have been used. Tests were conducted on samples similar to those used for trench tests, as well as on washed samples to discard the fines. This series of tests aimed at investigating the efficiency of hydro-compaction under more controlled conditions as well as to detect the role of fines content on the compactivity of natural sands.

Field Experiments

A normal practice of sewer line construction is to lay pipes at certain depth between manholes over a granular bedding layer, which is then covered, gently by sand layers of 150 and 200 mm thickness. This is to protect the pipe from the following mechanical compaction of the overlying successive sand layers of 200-300 mm thickness. This process takes a long time and costly effort; however it is the commonly followed procedure in engineering practice to achieve the required results of compaction. In some cases, especially when excavations are located in open areas, contractors recommend to use the procedure of compaction by watering, i.e. flooding the fill material in the trench with water or simply hydro-compaction. They claim it is easier and can achieve the result in less total time and effort. Generally and as per contract, it is the responsibility of the contractor to select the method of compaction provided he achieves the required specified degree of compaction. In the present case the contractor proposed to use hydro-compaction for the pipe trenches being easier and faster. The debate was whether the method is reliable and able to achieve the required compaction, with an acceptable degree of uniformity. It was agreed with the contractor to carry out experiments on the method of hydro-compaction before it can be approved as an acceptable procedure. Accordingly a set of field experiments was designed by the author to test the effectiveness of such a method using the same natural sand. Two trenches have been selected for experiments where moist fill sand was placed to a thickness of 1000 mm in the first one and 600 mm in the second, and then water was applied at the top of the fill sand until the trench is flooded by water to a height of about 200 mm. Field density tests have been carried out using the sand cone test method, after 2 duration periods of 2 and 18 hours since flooding the trench with water, in order to test the effect of time factor (Figure 4). The density tests were carried out at different lateral and vertical positions of the trench in order to test the homogeneity of compaction throughout the trench. Figure 5 shows a schematic diagram of the tested trenches indicating the position of density tests for both 2 hours and 18 hours test series.

Laboratory experiments

In order to study the effect of the fines content on the efficiency of hydro-compaction under more controlled conditions than in the case of field testing, laboratory experiments were designed and executed on the same natural sand that was used in the field experiments, however in those tests washed sand on sieve No.50 to get rid of the fines, have been used in addition to the normal natural sand. A 200-liter drum was used for the laboratory experiments.


Figure 4. Photographs for density testing on site, (a) near surface test, (b) deep test


Figure 5. Schematic diagram of the test trenches and the test locations

The drum was perforated from the bottom and the sides and then fixed on a gravel layer to allow good drainage of the sand, and then wrapped with a membrane to prevent the migration of sand particles during inundation with water (Figure 6). The drum was filled with known weight of sand in three layers to help getting homogenous samples of density of 1.6 MG/m3. Then water was applied to the drum and kept at a constant head of about 150 mm over the sand surface, for a period of an hour, during which draining water was observed from the sides and bottom of the drum. The amount of settlement of the sand surface as a result of inundation by water was recorded and therefore the consequent change of density was calculated. Also sand cone tests were carried out at two different depths (200 and 600 mm) from the surface of the drum both before and after inundation.


Figure 6. Photographs for drum testing, (a) during watering; (b) after watering


Field Experiment No. 1

The length of the test trench is about 30 m; a layer of sand was laid in the trench to a thickness of 1000 mm and then flooded with water gradually until water accumulates at the surface to a height of 200 mm, Presumably the water percolates downward through the sand layer down to the gravel bedding of the pipe. Two hours later, five cone density tests were conducted at a depth of 200 mm from the surface at distances (5, 10, 15, 20 and 25 m having 5 m distance apart) from the edge of the trench (Figure 5a). Tests gave results of 83.9%, 85.7%, 82.1%, 86.2%, and 82.9%, with moisture contents of 13.5%, 11.5%, 13.5%, 10.8% and 12.1% respectively (Table 1). Then other five cone density tests were carried out at depth of 600 mm from the surface to detect the compaction at depth. Those tests gave the results, 81.7%, 86%, 83.5%, 83.1%, and 107.5%, with moisture contents of 13.6%, 13.1%, 14%, 14.5%, and 14.6% respectively (Table 1). The results fail to meet the required compaction of 95%. A possible reason for that is the high moisture content, and therefore it was decided to retest the hydro-compacted sand after giving more time for drainage and possible better compaction. After 18 hours a second series of density tests were carried out both at the surface and at depth. Six tests were conducted at 200 mm depth from the surface, at distances of 2.5, 7.5, 12.5, 17.5, 22.5 and 27.5 m from the edge of the trench (in between the locations of the previous tests). The results gave higher density results of 88.5%, 89.7%, 87.8%, 81.9%, 86.5% and 86.4%, with generally, less moisture contents than the previous tests, of 10.4%, 10.5%, 10.2%, 13%, 10% and 10.4% respectively. Four other tests were conducted in the same horizontal locations but at depth of 600 mm. The results were 87.9%, 104.8%, 91.5%, 82.7%, 86.7% and 85%, with moisture contents of 11.5%, 15%, 10.5%, 12.4%, 11.1%, and 12% respectively. The later results show some improvement of compaction, however still could not reach the required compaction of 95%, except for a few tests.

Table 1. Test results of field density and moisture contents


Field Experiment No. 2

In order to investigate the influence of the thickness of the sand layer and to detect whether the hydro-compaction is more efficient on thinner sand layers, a second trench of similar dimensions was selected for testing; however the sand layer was laid to a thickness of 600 mm. The same procedure of the previous tests was followed and density tests were carried out for similar conditions. In this trench, the density tests were conducted at depths of 200 mm and 450 mm from the surface to suit the thinner layer, after the same durations of 2 and 18 hours, for the purpose of comparing the results of the two thicknesses having the same conditions. The results were 82.5%, 84%, 83.1%, 81.9% and 82.6% with moisture contents of 13%, 11.7%, 14.8%, 11.7% and 12.6% respectively for the surface tests, while the deep tests gave the results of 81.5%, 84.3%, 86%, 81.4% and 102.5%, with moisture contents of 13.4%, 12.5%, 11.5%, 13.5% and 14.1% respectively (Table 1). Another series of near surface and deep tests were carried out after 18 hours similar to the first trench. The results of the near surface tests were 88.5%, 86.2%, 89.8%, 90.3%, 87.8% and 91.5%, with moisture contents of 10.6%, 11.1%, 10%, 9.8% and 10.3% respectively (Table 1). However the results of the deep tests were 89.5%, 84.8%, 86.5%, 85.2%, 84.8% and 88.3%, with moisture contents of 10.85%, 12.4%, 11.3%, 11%, 11.8% and 10.1% respectively (Table 1).

Laboratory Experiments

Laboratory tests on the natural sand gave relative compaction results in the range of 88% to 93%, which shows slightly better compaction than in the trenches. This could be attributed to the smaller size of the samples, which allowed better circulation and drainage of water through the sand and therefore better packing by rearrangement of the sand particles. This further explains why the washed samples gave even better results, where the relative compaction ranged from 90% to 95%. These results confirm the effect of the fines content, even small percentages in the range of 5%, on the compactivity of the sand following hydro-compaction, although they have positive influence in case of the common practice procedures of compaction in which the fines fill the spaces between the larger particles. It seems that such an effect is masked by the role of drainage in case of hydro-compaction.


The site trial experiments on hydro-compaction of sand layers, showed variable percentages of compaction with ranges between 81% to about 90% relative to the maximum dry density obtained from Standard Proctor Compaction tests (Table 1). Three abnormal results of 102.5%, 104.5% and 107.5% were recorded, and these obviously indicate error of measurement, which is not uncommon in locations of high moisture content, where the possibility of slumping of the walls of the test holes exists. The tested samples showed a range of moisture content from 9.8% to 14.5%. The results generally indicate and as expected, sort of relationship between the attained percentage of compaction and the moisture content, where higher results coincide with moisture contents close to the optimum moisture content of the Proctor Test (Table 1 and Figure 7). The figure indicates that the general trend of the points follow a line which could match the right limb of the standard Proctor curve i.e beyond the optimum moisture content.


Figure 7. Relative compaction vs. moisture content in the test locations of the trenches

Density tests have been conducted after two soaking durations, 2 hours and 18 hours for the trench tests and 1 hour for the laboratory tests. Test results indicate that a certain degree of compaction up to 91.5% can be achieved by watering the sand, however not to the required degree for a structural fill. Figure 8 shows the frequency distribution of the test results, indicating that 3 test results were between 90% and 95% which the range of non-structural fills, 3 results were abnormal. About 40% of the results were in the range between 85% and 90%, which for certain conditions can be accepted for non-structural fill material. However, it is worth mentioning that the results are very inconsistent; as the measured relative compaction values are random both laterally and vertically (Figure 9). The figure also indicates that the range between near-surface and deep test results vary along the trench.

Figure 8. Frequency distribution of the field density test results

A main factor is the moisture content, which was found high, and this has prevented successful and accurate density measurements due to potential slumping of the density hole walls and this was evident by the abnormal results of higher than 100%. Another problem is the random distribution of moisture content as measured both laterally and vertically throughout the trench (Figure 10), ranging between 9.8% and 15%, with measured high moisture contents even after drainage period of 18 hours. This is a reflection of the water flow through the soil, which normally follow the paths relative to soil permeability that is directly influenced by the distribution of fines within the sand grains, though the percentage of fines is small.

The results generally indicate that the quantity of water added to the sand, the duration of soaking and the method of water application are controlling factors in this method of hydro-compaction, in addition to the existence of even small quantities of fines, which could exist in concentrations within the sand layer, and therefore cause zones of low permeability that divert the flow of water within the sand layer. This could explain the variations in moisture contents both laterally and vertically. Variations in water quantity and time can lead to random results as the trials have demonstrated. In some cases high moisture contents were recorded long after watering (about 18 hours) and this indicates that the drainage is not uniform through the trench and such non-uniformity is responsible for the heterogeneity of the density results.


(a) Trench 1

(b) Trench 2

Figure 9. Compaction test results in the test trenches

However, among the random distribution of test results, there are indications that the longer drainage period, the more chance to get higher relative compaction (Figure 9), with the tendency of the near surface tests to generally give higher relative compaction than the deep tests of the same drainage period, which could also be related to the possibility of better and relatively faster draining of the top part of the layer.

Laboratory tests indicate a slight improvement of compaction over the trench tests, though not very distinctive. This could be attributed to the smaller size of the samples allowed better circulation and drainage of water through the sand and therefore better packing by rearrangement of the sand particles. This further explains why the washed samples gave even better results, where the relative compaction ranged from 90% to 93% (Table 2). These results confirm the possible effect of the fines content, even small percentages in the range of 5%, on the compactivity of the sand following hydro-compaction, although they have positive influence in case of the common practice procedures of compaction in which the fines fill the spaces between the larger particles. It seems that such an effect is masked by the role of drainage in case of hydro-compaction.

Figure 10. Distribution of moisture contents in the test trenches (a) Trench 1 and (b) Trench 2


1. Measured moisture content and density values of hydro-compacted sand were random both laterally and vertically, and no pattern was detected in the spread of results. Therefore hydro-compaction is considered inadequate to give uniform compaction to high relative densities of greater than 95%, that is required for structural fills.

2. The random results demonstrate beyond doubt that one positive result does not indicate consistently good compaction. Therefore quality control is very difficult and consequently there is great risk of settlement of the fill.

3. Hydro-compaction is more efficient on washed sand with less or no fines content that can be better drained, however its not practical and costy to use washed sand for backfilling.

4. Hydro-compaction requires huge quantities of water which can be detrimental to adjacent structures and underground services.

5. Hydro-compaction is generally slow process. Site trials on thick layers have demonstrated that it has no advantage over the conventional mechanical compaction in terms of speeding up the process of backfilling. In fact mechanical compaction of thin layers with proper management and enough equipment used by skilled labors can be faster and above all, conventional compaction ensures that compaction can be seen and confidently approved after testing.

6. The percentage of failures was high, 35% in 600 nun layers and 75% in the 800 mm layers. Also inaccurate results were obtained from inadequate density test procedure and high moisture content.


  1. Anon. (1976) Methods of Tests for Soils for Civil Engineering Purposes, BS 1377, British Standards Institution, London.
  2. ASTM (1990) American Society for Testing and Materials. Standard D - 1557: Test Method for moisture - density relations of soils and soil - aggregate mixtures using 10 lb (40.54 kg) rammer and 18 in (457 mm) drop. Annual Book of ASTM Standard, Vol. 04.08.
  3. Bagdasarov, Yu. A, Rabinovich, I. G., Epaneshnikov, L. O. and Khanzhin N. A. (1992) Compaction of collapsible soils by soaking with tamping during the collapse process. Soil Mechanic and Foundation Engineering, Vol. 28 No. 4, P 159 - 164.
  4. Bell, F. G (1993) Engineering treatment of soils. E & IN Spon, an imprint of Chapman & Hall, London, PP 302.
  5. BS 1377 (1975) British Standards Institution. Methods of test for soils for civil engineering purposes. Test No. 15 (B).
  6. National Clay Pipe Institution (1990) Clay Pipe Engineering Manual. Santafi Springs, California, USA, PP 114.
  7. Parsons, A. W. (1987) Shallow compaction, in Ground Engineer's Reference Book (ed. F. G. Bell), Butterworths, London, P. 37/13 - 37/17.
  8. ASTM (1990) American Society for Testing and Materials. Standard D-698: Test Methods for Moisture - Density Relations of soils and soil aggregate mixtures using 5.5 lb (20.4 kg) rammer and 12 in (304.8 mm) drop. Annual Book of ASTM standards, Vol. 04.08.
  9. Duncan, J. (1992) Soils and foundations for architects and engineers. Van Nostrand Reinhold, PP 361.
  10. Watson, I and Burnett, A. D. (1995) Hydrology An Environmental approach. Ft. Lauderdale: Buchanan Books, Cambridge, PP 702

    © 2004 ejge