Chemical improvement of soluble rocks: Foundation of Mosul Dam as case study

The dissolution of soluble rocks (gypsum/anhydrite) beneath the Mosul Dam by water seepage has been observed upon the initial impoundment; consequently, several sinkholes have been manifested in the vicinity of the dam site. Traditional grouting has been envisaged as a potential remedy; however this measure has not eradicated the problem. The main purpose of this study is to overcome the solubility of the gypsum/anhydrite rocks using chemical grouts. Rock samples were acquired from the Fatha Formation outcrop and problematic layers of brecciated gypsum situated at varying depths beneath the Mosul Dam. Two commercially available liquid polymers, polyurethane (PU) and a mixture of acrylic and cement (ARC) were used to investigate their sealing performance in halting of the solubility of the rocks (gypsum/anhydrite). To simulate the dissolution phenomenon under the influence of artificial hydraulic pressure of the dam and the water flow in its abutments, two distinct laboratory models were devised. The outcomes from the experimental study on both untreated and treated samples revealed that the acrylic-cement composite (ARC) and polyurethane (PU) are influential polymers in halting the solubility of the gypsum rock samples under both factors of water pressure and high-velocity water flow.


Short background on chemical grouting
Chemical grouts were created in the early 1900s to reinforce and control water flow in geological units with small pores which are unsuitable to be grouted by regular cement grouts.Chemical grouting entered its contemporary phase in the early 1950s, propelled by advancements in polymer industries 22 .Polymers are generated by chemically reacting monomers which are small molecular compounds.They typically consist of multiple structural units joined together by covalent bonds 23 .
Grouts may be categorized as the traditional (i.e., cement, fly ash, lime, bitumen, etc.) and non-traditional (e.g., polyurethane, acrylate, sodium silicate) 24 .The main advantages of the polymer grouts are their high penetrability, low viscosity, controllability of their gelling time and high adhesivity.These characteristics render them to consider as viable alternatives to traditional cement 25,26 .In recent years, chemical grouting materials such as polyurethane and silica sol 27 , butyl acrylate and styrene 28 , sodium silicate 29 , and polyacrylamide 30 have been utilized in soil remediation.
PU is the product of combination of polyol (-OH) and isocyanate (-NCO).Depending on its response to water, PU is classified into hydrophobic and hydrophilic.Hydrophobic absorb minimal additional water, while, hydrophilic grouts assimilate considerable amounts of water into their chemical structure 31 .
Acrylic polymers or thin film coatings represent another category of polymers.In recent years, the use of thin film coatings has jumped into the engineering field, particularly in soil and rock stabilization 38 , owing to their broad range of features like flexibility and excellent adhesion, mechanical strength, scratch resistance, and thermal stability 39,40 .
In their study, Chhun et al. 40 have concluded that the acrylate-cement grout played a crucial role in enhancing the mechanical strength of silty sand.Kolay et al. 26 have investigated the impact of liquid acrylic polymer on geotechnical properties of fine grained soils.They have noted that the utilized grout does not offer a significant impact on the Atterberg limits and unconfined compressive strength; however, it influences the CBR results.

Mosul Dam, its foundation, and a brief history of sinkholes development
Mosul Dam is a multipurpose earth fill embankment located on the Tigris River in northwest of Mosul city in northern Iraq. Figure 1 shows the Mosul Dam location.This hydraulic structure with a storage capacity of 11.11 billion m 3 started operating in 1986, and from then on, it struggled with seepage along with the dissolution of (1) the gypsum beds in the dam foundation and its abutments 3,41 .From structural geology standpoint, this project is situated between two anticlines, Butmah west anticline on the right abutment and the Taira anticline on the left abutment 1,21 .Two prominent geological formations exist at the dam site and reservoir, including Fatha Formation (Middle Miocene) with lithology of marl, limestone, and gypsum sequences, and the Euphrates-Jeribe Formation (lower Miocene) comprising limestone, dolostone beds, and marl 1, 3,42,43 .Figure 2 represents the schematic diagram of the dam cross section and its foundation.Due to the presence of soluble rocks, karstification is the most significant geological phenomenon at the Mosul Dam site.Sinkholes which have developed in both the dam site and the reservoir area are the most prevalent feature of this phenomenon.The existence of brecciated gypsum (GB) at different depths of the dam foundation is another feature [44][45][46] .www.nature.com/scientificreports/ In 1986, a number of sinkholes were recognized on the right bank, approximately 150 m away from the dam contact with the right abutment 41 .Between 1992 and 1998, four sinkholes in a linear arrangement parallel to the dam axis developed 1 .Also, in February 2002, a large sinkhole with 15 m depth and width appeared just 150 m downstream of the dam toe at the left bank 21 .These sinkholes prove continuous gypsum dissolution in the dam foundation and abutments.

Gypsum rocks and sampling
In this research, the sulfate rock samples were collected from two sources: the Fatha Formation outcrop or (surface samples-S) and boreholes within problematic layers of the Mosul Dam foundation (G).Surface samples were collected in large pieces and then cut into blocks (Length = 25 cm, Width = 15 cm, variable height of 4-6 cm) for dissolution test under high-velocity water flow, simulating surface dissolution in the dam abutments Fig. 3a.The core samples from boreholes Fig. 3b, labeled as G1 (from GB1, depth of 44m), G2 (from GB0, depth of 76 m), G3 (from GB3, depth of 94 m), and G4 (from GB2, depth of 130 m) were retrieved from the Mosul Dam Core Samples Conservation Warehouse.These core samples with 5 cm diameter were trimmed to achieve a length-to-diameter ratio of 2.3.They were then employed to quantify their solubility under pressurized water flow (400 kPa), simulating rock dissolution in the dam foundation.
The remaining fragments from each sample were reused for their chemical analysis.To identify the mineral composition of the rock samples, two methods were employed: microscopic analysis (thin sections) and X-ray diffraction (XRD).
The microscopic analysis revealed that the predominant components of both type rock samples were gypsum and anhydrite.
On the other hand, X-ray diffraction (XRD) analysis revealed that gypsum with 2ϴ peak of 11.6, 20.65, 23.28 and 29.05 constitutes the primary component of the surface specimens and G2 Koukouzas and Vasilatos 47 achieved a similar result.While G3 and G4 samples were recognized as anhydritewith 2ϴ peak of 25.45 and accessory minerals of calcite with 2ϴ peak of 40, 43 and 48.A similar result was achieved by Al-Jaroudi a 48 .The G1 sample exhibited a mixture of both gypsum and anhydrite with 2ϴ peak of 25.5 and 55.5, respectively.Figure 4 shows the microscopic images and XRD patterns of the samples.

Acrylic-cement grout (ARC)
The emulsion liquid acrylic polymer utilized in this study was mixed with type II Portland cement and aluminum sulfate accelerator (1.2 Molarity) in different proportions of (Acrylic: Cement: Al 2 SO 4 by weight = 5:0.5:0.5, 5:0.75:0.5, 5:1:0.5 and 5:1.5:0.5).The cement contents of the grout were varied, while the polymer weight and the aluminum sulfate proportions were kept constant.Choosing a 0.5 ratio of aluminium sulfate (1.2 M) was based on the consideration that more than that ratio would have lowered the initial fluidity of the grout to a great extent.The physical and chemical specifications of the liquid polymer are presented in Table 1.

Polyurethane (PU)
In this study, the hydrophobic foam (Seal Boss 1510) based on diphenylmethane diisocyanate MDI polyurethane was utilized.Its gelling time was adjusted by incorporating a 15 × accelerator.The physical and chemical specifications of the PU are given in Table 1.
The reaction of the foam varies based on the accelerator quantity.The higher accelerator percentages lead to increased reactivity, shorter hardening time, more expansion, and lower density, this was confirmed by 49 .To achieve the polyurethane foam with different densities and gelling times, the range of accelerator quantities were varied with 5 percent interval from 5 to 25%.

Sample treatment
The treatment procedure of the samples with ARC is represented in the Fig. 5a.In the treatment of the surface samples, all surfaces of the sample were coated with ARC.At first, all surfaces of the sample except one surface were coated (painted) by the ARC grout.This created a uniform layer with 2 mm on the coated surfaces.After setting this layer, the last surface of the sample was coated with 2 mm thick layer of ARC.Finally, the coated sample was left at laboratory temperature (25 ± 2 °C) for 24 h and then subjected to a dissolution test at 0.3 m/ sec water flow velocity.
For the treatment of the core samples with ARC, initially, the surfaces of the sample along its height as well as one of its ends were coated with a uniform layer of the ARC.This caused an approximate 4 mm increase in the sample diameter and 2 mm in its height once the ARC had solidified (a caliper was used to measure the diameter of the treated sample).After drying this part, the other end of the core sample was coated with 2 mm thick layer of ARC; thus, the sample's height increased by 4 mm.After setting of the last part, the treated core sample was left for 24 h in the room temperature.Finally, the solubility of the ARC treated core samples were investigated under constant pressure of 400 kPa.
Concerning the treatment of the samples with PU Fig. 5b, the same approach used for ARC treated samples was used to treat the samples with PU.However, due to the variation of PU expansiveness with the change in accelerator amount, it was slightly challenging to treat the samples with PU.Therefore, the PU thickness on the  www.nature.com/scientificreports/surface of the treated samples for different tests with different accelerator ratio was not uniform.PU with a greater accelerator ratio is more expansive and lower density.The total number of PU treated samples for the surface (S) and core samples (G) were 10.Five samples of each type were treated with different accelerator ratio (S-5% acc to S-25% acc) and (G2-5% acc to G2-25% acc).
It is worth mentioning that the S-treated samples were tested under high-velocity water flow and the G2-treated samples were tested under a pressure of 400 kPa, and all experiments were performed at a temperature of 25 ± 2 °C.

Dissolution test using velocity-base apparatus
This device was designed to simulate the conditions of the gypsum rock layers under water seepage with high velocity (discharge) in the dam abutments.Figure 6 represents the schematic of the velocity-base dissolution test apparatus.It comprises a Plexiglas container, a pump unit with associated parts, and a 120-L storage tank.
The Plexiglas container with a 76 cm length, 18 cm width, and 14 cm height is the testing vessel of the sample.The inlet and outlet of this container were inclined to minimize the turbulence flow of circulated water during the test.Additionally, a perforated plate was placed in both.The container has a Plexiglas cover with two inlet and outlet valves for air bubble removal, secured by a steel frame for integrity during the tests.To maintain stable temperatures throughout the test, the pump was separated from the electromotor.There is a perforated plate in the tank that prevents air bubbles during water circulation in the system.In this apparatus, the flow rate was measured using the V-notch method 50 .
Regarding the test preparation, firstly, a layer of pure sand was placed at the container base to allow the sample to reach the appropriate level at which the circulated water flows over its top surface.Next, the block of surface sample was placed on this sand layer.To ensure only the top surface of the sample was exposed to water flow, the sections touching the container inner wall were sealed with 1 cm thick silica glue.This prevented water leakage from coming into contact with the other surfaces beyond the top surface.Finally, the container was tightly covered, and 100 L of water were added to the tank.Now the sample was ready to be tested.
Using this apparatus, three untreated surface samples (S) were tested at different water flow rates of (0.07 m/s, 0.25 m/sec and 0.3 m/sec) and the treated surface samples (S-ACR and S-PU) were tested at 0.3 m/sec water flow rate (the highest flow velocity of the study), and at 25 ± 2 °C.

Dissolution test using pressure-base apparatus
To simulate the conditions of gypsum rock layers under water pressure at the dam foundation, this apparatus was designed.The main units of this apparatus are: (a) a Plexiglas cylindrical vessel with a height of 30 cm and a www.nature.com/scientificreports/constant internal diameter of 11 cm (b) a pump unit (c) a tank with a storage capacity of 250 L. Figure 7 illustrates the pressure-base dissolution test apparatus.As can be seen from the figure, at either end of the Plexiglas cylinder, there is a steel bench joined by three steel clamps, and screwed to prevent water leakage during testing.Each bench has its valve (inlet and outlet), which allow water circulation in the system, and an air vent is connected to the inlet valve to eliminate air bubbles.The configuration also incorporates a pressure gauge to monitor the testing pressure.The pump circulates the water between the tank and the Plexiglas cylinder vessel during the tests.
Regarding the test procedure, initially, all valves were closed.Subsequently, a 5 cm thick layer of gravel was placed at the bottom of the cylindrical container.A steel mesh was then placed over the gravel layer.After that, the core sample was positioned on this steel mesh.To remain the sample stable during the test, a grid was installed in the center of the sample.Next, another steel mesh plate was placed above the core sample, and a final layer of gravel 12 cm thick was added over the last steel mesh.Following this, the Plexiglas cylindrical vessel was secured using clamps and screws.The test preparation was finalized by adding 100 L of water to the tank.
The experimental steps started by opening the valves, followed by the incremental introduction of water at a subdued pressure from the tank into the container via the inlet valve until the container was entirely filled with water.Following this step, and verifying the lack of trapped air, the pump circulation rate was gradually raised through an inverter until the pressure of 400 kPa was gained.This setup enabled the pump to convey water from the tank into the container and subsequently circulate it back to the tank.
In this experiment, the untreated core samples labeled as G1, G2, G3, and G4 were tested for 48 h at a temperature of 25 ± 2 °C.Then, the sample with the highest dissolution rate (G2) was chosen for coating with grouts and further tests.
It is worth mentioning that, during the experiments, the water samples were collected at regular time interval to measure the concentration of dissolved gypsum in circulated water.The titration approach and inductively coupled plasma (ICP) were applied in the cases of untreated and treated samples, respectively.

Other tests
Electrical conductivity (EC) of the water samples was measured using ATC (Automatic Temperature Compensation) equipped device.
In the dissolution tests on untreated rock samples, the EDTA titration method proposed by Horvai et al. 51 was used to measure the dissolved gypsum content in the water samples.In this procedure, 50 ml of the water sample was measured with a pipette and poured into a clean conical flask.Then, 10 ml of a buffer (1 M NH4OH) solution was added to the flask to reach a pH of 10-11, followed by the addition of a Eriochrome Black T indicator.
The titration began with the gradual addition of the standardized EDTA (0.01 M Ethylenediaminetetraacetic acid) solution from a burette to the sample solution in the conical flask.The titration reached its endpoint as the solution changed color from red to blue, indicating the reaction between EDTA and calcium ions.The recorded EDTA volume was utilized to calculate the quantity of dissolved calcium using Eq. ( 4).To improve result accuracy, the test was done three times, and the mean of the results was taken.
where, E is the volume of EDTA and M is the molarity of EDTA (0.01).
In the dissolution tests on treated rock samples, Inductively Coupled Plasma (ICP) Spectroscopy was utilized to measure the calcium concentration in the water samples.This analytical test is utilized to measure and identify elements within a sample by assessing the ionization of elements present in the sample.

Results and discussion
Dissolution tests of the untreated surface samples (S) using velocity-base apparatus Three untreated surface samples with identical chemical composition at different water velocities of (0.07 m/sec, 0.25 m/sec and 0.30 m/sec) were tested until reaching the saturation point of the solvent (2.8 kg/m 3 ).Figure 8 shows the relationship between the quantities of dissolved gypsum versus time for the different water velocities.
It can be seen that all samples showed a similar increase in dissolved gypsum quantities over time, however, there was a noticeable difference in the saturation duration, reflecting the influence of different water velocities.Specifically, at a water velocity of 0.07 m/sec, the saturation point was achieved at 672 h; however, when the water velocities were 0.25 m/sec and 0.30 m/sec, the saturation duration reached 576 h and 528 h, respectively.This finding aligns with the results of Imam et al. 52 .
By utilizing Eq. ( 3), drawing a graphical relationship between Cs Cs-C and A v' t , and determining the dissolution rate constant for each test, the correlation between dissolution rate and water flow velocity was further clarified.Figure 9 shows the effect of water flow velocity on dissolution rate constant.
As can be seen from Fig. 9a, when the velocity of circulated water was 0.07 m/sec, the value of K was 5.78 × 10 -6 , while the value of K reached 7.8 × 10 -6 and 8.5 × 10 -6 at the velocities of 0.25 m/sec and 0.30 m/sec, respectively.The variation of K was plotted as a function of water velocity in Fig. 9b.It can be noted that there is a linear relationship between K and V. Similar results were reported by James and Lupton 15 .
It is noteworthy that the velocity with the highest dissolution rate (0.3 m/sec) was used in testing the treated samples with ARC and PU.
The electrical conductivity (EC) of tap water and saturated water samples were 800 μS/cm to 2150 μS/cm, respectively.This technique was used as an index test before titration and ICP tests of the water samples.Minor variations in the EC results were due to the presence of other ions in the water samples.Figure 10 shows the calcium (gypsum) concentration and EC relationship for the untreated sample tested under 0.3 m/s water flow.

Dissolution tests of the core samples (G) using pressure-base apparatus
Four untreated core samples (G1 to G4) with different chemical compositions (gypsum and anhydrite) were tested under a constant pressure of 400 kPa for 48 h. Figure 11 demonstrates the graphical relationship between the quantities of dissolved gypsum over time.
It can be observed that G2 has displayed the most rapid dissolution rate, whereas G4 showed the minimum dissolution rate throughout the testing period.This distinction can be attributed to variations in the chemical composition of the respective samples.Based on the chemical analysis finds, G2 is predominantly composed of gypsum, while G4 is primarily constituted of anhydrite.This finding agrees with the result of Zanbak and Arthur 53 , they concluded that under water pressure conditions, anhydrite exhibits lower solubility compared to gypsum.
Notably, the sudden changes in the slopes of the curves for sample G4 at t = 40h, and G2 at t = 17h may be attributed to the weakened effect of the water on the structural integrity of the samples over time, consequently, the samples have experienced surface or internal fractures or structural weaknesses that accelerated dissolution rate.
It is noteworthy that among the tested core samples, the sample with the highest dissolution rate (G2) was selected for additional study after being treated with ARC and PU.

Dissolution test of the ARC treated gypsum rock samples
Fig. 12 illustrates the effect of ARC on the gypsum rock solubility at the highest flow velocity of the study (i.e., V = 0.3 m/s) (Fig. 12a), and under 400 kPa water pressure (Fig. 12b).www.nature.com/scientificreports/Portland cement.Also, the cracking phenomenon in grouts due to the content of more than the optimal ratio of cement has been confirmed by Akinyemi and Omoniyi 56 .On the other hand, the starting of the dissolution process in the (5:1.5 ratio) ARC treated sample showed the resistance of the ARC grout to hydrolysis, in other words, it revealed the influence of both factors of water velocity and pressure on the grout.This result aligns with the outcome obtained by Momber and Kovacevic 59 ; They concluded that high-velocity water flow leads to erosion of concrete, which is primarily driven by the formation and propagation of microcracks.ARC with different cement contents from 0.5 to 1.0 percent has effectively hindered the solubility of gypsum over a given duration.The variation in these ratios impacts the setting time of the grouts.The selection of the most suitable percentage of acrylic to cement is contingent upon the target of the grouting.

Dissolution test of the PU treated gypsum rock samples
Fig. 13 compares the results of a series of dissolution tests conducted on both untreated and PU treated samples at the highest flow velocity of the study (i.e., V = 0.3 m/s) (Fig. 13a), and under 400 kPa water pressure (Fig. 13b).As evident from the results, PU demonstrates a substantial positive effect on halting the gypsum sample solubility.
As the untreated gypsum curve exhibited a consistent upward trend in dissolution, the curve representing the PU coated samples remained stable at the original concentration (tap water salinity) for all various PU/acc ratios (Due to the similarity of the treated samples results, the curves of the different tests are shown as a single curve in the Fig. 13a,b.Furthermore, although the density of the grout decreases with increasing accelerator content, and consequently the expansibility of the gelled PU increases, interestingly, test results for all accelerator contents (from 5 to 25% accelerator) showed similar outcomes, which was the prevention of the gypsum dissolution  during the testing period.This can be attributed to the strong resistance of the PU to hydrolysis and erosion.Similar outcome was obtained by Jacob 37 , who concluded that the durability of polymer adhesives in the presence of water is attributed to its high resistance to hydrolysis.Additionally, the used PU in this research represents the hydrophobic type of PU, according to the studies by Vipulanandan et al. 31 and Buzzi et al. 60 , hydrophobic materials on the foam surface create a water-repellent barrier, and significantly prevent water absorption due to the combination of surface repellency and the closed-cell structure (the cells are not interconnected).Therefore, water is unable to permeate the foam, and the closed-cell design minimizes the total moisture content.
Furthermore, the high resistance of PU against both water pressure and high velocity water flow, simultaneously the lack of gap between PU and the rock sample during the whole time of the test further indicates the strong adhesion between PU (with different density foams) and the gypsum samples.This result is consistent with the result of Saunders and Frisch 61 , who concluded that PU is a coating material with excellent adhesion to various substances.
When their waterproof capacity of the materials is compared, as seen fromFig.14, the PU is more impactful grout compared to the ARC.PU foam with different densities (different accelerator dosage) completely stopped the dissolution process under both factors.In contrast, ARC was able to achieve the similar result, but only when optimized at specific ratios of (5: 1.0).
Nevertheless, there exist essential distinctions between the two types of grouts.ARC lacks a quick setting, whereas the gelling time of PU is quicker and to some extent controllable.additionally, compared to ARC, PU is more penetrable and can fill much smaller cracks 25,26 , however it is more difficult to use underground because its behavior (expansibility and quick gelling time) cannot be controlled underground 60 .While ARC is more stable in different environments, nevertheless, there are defects with ARC utilization, particularly in its capacity to penetrate very fine soils or reach the narrow cracks and fissures within rocks 62 .On the other hand, in terms of cost, ARC is more economical compared to PU.

Conclusions
The purpose of the current study was to overcome the solubility of gypsum rocks using chemical grouts.The used rock samples were taken from the surface gypsum rocks and problematic layers of brecciated gypsum situated at varying depths of the Mosul Dam foundation.In summary, the main results derived from this simulated experimental investigation can be summarized as follows: •Under both factors of water velocity and pressure, Except for the case involving 1.5 percent of cement, the (ARC) grout has a significant effect in controlling the dissolution of soluble rocks, and plays a vital role in attaining full hydrolysis resistance under the influence of both water velocity and water pressure factors.Beyond the optimal ratio of 5:1, with increasing cement content the sealing strength of the ARC grout decreases, indicating that using cementitious grout alone cannot yield satisfactory results in treating gypsum rocks.This reveals the defect in the protective measurement of the Mosul Dam foundation, due to the use of cement and certain additives during the grouting.It also validates the source of persistent seepage at the Mosul Dam.
•The water pressure resistance of ARC is lower in comparison to its resistance against water flow velocity.The sample coated with ARC at a ratio of 5:1.5 experienced dissolution after 336 h under high-velocity water flow and after 168 h under pressure of 400 kPa.•From a perspective of sealing strength, PU proves to be a more effective choice than ARC.This is evident in the PU treated gypsum samples; they did not experience any dissolution during the given time, whereas the sample coated with acrylic/cement ratio of 5:1.5 exhibited dissolution during the experiments.

Figure 2 .
Figure 2. Schematic representation of the geometry of the Mosul Dam and the stratigraphic column of its foundation.

Figure 3 .
Figure 3. Gypsum rock samples, (a) blocks of surface rocks, (b) core samples taken from boreholes.

Figure 4 .
Figure 4. Microscopic images and corresponding XRD analysis of the surface and core samples.

Figure 5 .
Figure 5. Schematic illustration of the sample treatment procedure, (a) ARC and (b) PU.

Figure 6 .
Figure 6.Schematic illustration of the velocity-base dissolution test apparatus.

Figure 8 .
Figure 8. Variations of gypsum concentration over time in dissolution test of the untreated surface samples at different flow velocities.

Figure 11 .
Figure 11.Variation of quantities of dissolved gypsum over 48 h for untreated borehole core samples under constant pressure of 400 kPa.

Figure 12 .
Figure 12.Effect of ARC grouts on the solubility of gypsum rock samples with respect to different cement contents, (a) surface samples under 0.3 m/sec flow water rate, (b) core samples under 400 kPa pressure.

Figure 13 .
Figure 13.Dissolution test for the untreated and the polyurethane (PU) coated gypsum samples with respect to various accelerator additions (different density foams), (a) Surface samples under 0.3 m/sec flow water rate, (b) core samples under 400 kPa pressure.

Table 1 .
Physical and chemical properties of acrylic and PU.