Journal of Materials Science and Engineering A 12 (10-12) (2022) 115-130

doi: 10.17265/2161-6213/2022.10-12.002
D DAVID PUBLISHING

Moroccan Phosphogypsum Use in Road Engineering:

Materials and Structure Optimization

Diouri Ayad Chaimaâ1, Lahlou Khaled1, Alaoui Amina2 and El Omari Kamal3
1. Ecole Hassania des Travaux Publics (EHTP), Oasis, Casablanca 81081, Morocco
2. OCP S.A., Paris Office 19, rue de l’Amiral Hamelin, Paris 75116, France
3. OCP S.A., 2-4 Al Abtal Street, Hay Arraha, Casablanca 20200, Morocco

Abstract: Morocco produces annually large quantities of phosphogypsum (PG), which permanent stacking rises technical and
environmental constraints. However, the valorization of this coproduct in civil engineering and especially in roads building, is a
promising solution within a circular economy frame. A first experimental one-kilometer-long pilot, incorporating four different PG
based formulations with a 7% cement addition, built at Safi in 2017, allowed positive mechanical assessment (by deflector) and
environmental one (by leaching test). In order to further evaluate PG use as road material, our experimental approach focuses here on
optimizing material mixtures—made of phosphogypsum (maximum content desired) treated with cement (to be minimized so as to
reduce the cost) and sand or steel slag as granular corrector—to meet mechanical requirements of a road base material. We first
identified and characterized phosphogypsum produced at the Jorf Lasfar plant and other materials used. Design of experiment is used
for modeling desired physical and mechanical responses and to establish domains meeting the required criteria for using the mixture
material either as road subgrade layer or foundation layer. In addition, through a parametric study, we evaluated the effects of traffic
level, soil bearing capacity and mechanical performance of treated phosphogypsum mixtures on pavement design for three different
pavement structures (mixed, reverse and structure with treated sub-base) and determined that the best to adopt for maximizing PG
recycling is the pavement structure with subbase treated with hydraulic binder.

Key words: PG, circular economy, road material, optimized mixture.

Nomenclature SS Steel slag

Diametral tensile strength at 28, 90 and
Ts28d, 90d, 360d
C Cement 360 days
Californian-bearing ratio after 4 days
CBR-4d
of water immersion 1. Introduction
Simple compressive strength at 7, 28
Cs7d, Cs28d, Cs90d
and 90 days
Phosphogypsum (PG) is a coproduct from natural
DOE Design of experiment phosphate rock processing, into phosphoric acid using
E28d, E90d, E360d Elastic modulus at 28, 90 and 360 days wet industrial process. The phosphoric acid wet chemical
The vertical swelling after 4 days of treatment process, or “Wet Process”, is widely used to
Gv
water immersion with overload
produce phosphoric acid (up to 90%) and calcium
RHB Road Hydraulic Binder
sulfate—mainly in the dihydrate form (CaSO4·2H2O).
IBI Immediate bearing index
The reaction is as follows (Eq. (1)):
International Fertilizer Industry
IFA Ca5F(PO4)3 + 5H2SO4 + 10H2O 
Association
(1)
OCP OCP Group S.A. 3H3PO4 +5CaSO4·2H2O + HF
PG Phosphogypsum The wet process is economical but generates a large
S Sand amount of PG (5 tons of PG per ton of phosphoric acid

Corresponding author: Diouri Ayad Chaimaâ, civil engineer

and graduate researcher, research fields: building materials and
structures.
116 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

produced) [1]. Thus, the world production of PG is PG was used as a binder to stabilize soil in place, in
estimated between 100 and 280 Mt per year [2]. mixtures showing more advantages than those with
Knowing physio-chemical properties of PG is of clay. Indeed, PG mixtures, after compaction, did not
great importance in order to develop technical solutions absorb large amounts of water, thus minimized delays
for the management and/or sustainable use of this due to rainy days. Shrinkage and swelling cracks were
coproduct. PG properties depend on the nature of the also reduced, with a gain in stability and pavement base
phosphate ore, the wet process used and operating strengthening over time. For example, a control section
efficiency of the plant. with PG built in Florida has proven to show better
In terms of chemical composition, as the main mechanical properties than the one built with
component of PG is gypsum, so sulfate and calcium can conventional materials, after 21 years of use. In Texas,
constitute more than 90% of its chemical composition. since 1991, test sections in city streets, shopping
Wet process or inheritance from the ore could explain centers, truck terminals, parking lots and loading
presence of other elements in PG. platforms have been successfully constructed using
From a morphological point of view, PG has, cement-fly ash stabilized PG.
depending on phosphate rock source, a majority of  In Russia, in 2014 [6], firstly in Moscow and then
particle size between 0.045 mm and 0.250 in diameter in Rostov region, level of measured radioactivity in the
[3]. PG has a crystalline structure, mainly of rhombic experimental roads showed there is no local restriction
and hexagonal shapes. on PG use in road construction.
In terms of physical properties, due to acidic phosphoric,  In Finland [6, 7] on a rural road in Maanikan, PG-
sulfuric and hydrofluoric residues contained in the fly ash mixtures worked perfectly in comparison with
porous PG, it is considered an acidic coproduct (pH < conventional materials. PG use, as an industrial
3): this point is to be considered for any further use. The coproduct, has shown to be environmentally unharmful,
solubility of PG depends on its pH, and PG is very as laboratory tests indicate, but more significantly as
soluble in salt water (4.1 g/L). Its particle density is proven through experimental pilots.
between 2.27 and 2.40 g/cm3 [4] and its bulk density is In 2018, in India [8], phosphogypsum from Paradeep
between 0.9 and 1.7 g/cm3 [5]. The vertical hydraulic was used in backfill and subgrade layers. Other works
conductivity of phosphogypsum would be between 1 × followed in China; in 2020 [9], for example,
10-5 and 2 × 10-7 m/s [4]. researchers tried to establish pH and particle fineness
Road construction, which consumes large quantities effects on mechanical properties of PG mixtures for
of granular materials, appears to be one of the practical stabilized soil-cement-PG for road construction use.
and promising valorization ways for PG. Indeed, Researches were led in Morocco too, as in 2020 [10],
recycling this type of material could provide interesting an evaluation of raw phosphogypsum use alone as
solutions to current construction materials shortage, embankment material and stabilizing effect of various
within a circular economy framework. However, this additives as clayey soil, fly ash, lime, calcareous
objective needs a rigorous scientific methodology and material and a special hydraulic road binder, in order to
pilot experiments to demonstrate its relevance. use these mixtures as a base material in road pavement,
The history of PG’s use in roads building around the while ensuring a neutralization of PG acidity. Another
world presented by IFA [6], proved this valorization feasibility study in 2021 focused on the stabilizing
way meets specific local needs, preserves raw building effect of fly ash and lime addition on the strength and
materials, and has a negligible environmental impact: durability of PG use as road material [11]. These
 USA [6] was the first to experiment it since 1985, research works are exclusively laboratory tests.
 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 117

In Morocco, OCP Group, leading global producer of the frame of circular economy. This is why we tested
phosphates and their derivatives, processes phosphate steel slag, another local coproduct of the metallurgical
rock into phosphoric acid using the wet process at Safi industry, as granular corrector. In addition, in order to
and Jorf Lasfar industrial complexes, which generates optimize cost, we varied the cement content. To
significant quantity of PG. Several valorization ways complete the optimization of the materials, we also
are being explored for this co-product, among which conducted a parametric study to evaluate different
PG use as road material. So, in 2013, OCP group launched pavement structures according to the bearing capacity
a scientific partnership with academic institutions— of the soils in place, the level of the traffic and the
ENPC (Ecole Nationale des Ponts et Chaussées) and mechanical properties of the PG-based mixtures.
EHTP (Ecole Hassania des Travaux Publics)—and In the first section of this paper, we carried out a
professionals—the CNER (Centre National des Etudes detailed chemical characterization of Jorf Lasfar’s PG
et de Recherches Routières) at Ministère de l’Equipement, and all the other materials used in this study as granular
du Transport, de la Logistique et de l’Eau (Ministry of corrector (steel slag and crushed sand) or hydraulic
Equipment, Transport, Logistics and Fermer la parenthèse1 binder (cement). We define mechanical criteria for a
and GTR (Colas subsidiary in Morocco), to lead research material use in road construction and explain why raw
works and then develop pilot road sections at Safi site PG cannot be used alone neither in subgrade or
in 2017 then at Jorf Lasfar industrial complex in 2019. pavement layer, then we detail all factors and responses
The first experimental pilot road built in Safi region with the experimental methods used. Then we present
used as subgrade layer four different PG mixes, with methods for material optimization by varying the
either phosphate washing tailings or sand as granular proportions of the constituents: a mixture design approach
corrector and a fixed addition of cement (7%), based on (DOE) allowing determining mathematical modeling
literature review [12]. The mechanical assessment, of needed physical and mechanical properties of
carried out using Lacroix deflectograph after two years phosphogypsum mixtures. The determination of the
of heavy traffic, proved that PG mixtures gave better mechanical properties requires prior knowledge of the
results than the control section made up with soil in optimum water content in order to determine the
place. In fact, the elastic modulus gave better long-term density of the mixture. We present the method we have
bearing capacity classes for the four sections (PF3 and developed for this purpose. To evaluate the impact of
PF4) than for the control one (PF2) [13], with a service pavement structure, and assess which pavement structure
lifetime of 25 years. Moreover, environmental would be the most relevant one to recycle the maximum
assessment, based on leaching tests, showed that all amount of phosphogypsum at the lowest cost, a parametric
measured traces elements concentrations, were well study [14], using numerical simulations with Alize-
below national and international limits. LCPC [15]design model analyzes the effects of traffic
Based on these good results, the objective of the level, soil bearing capacity and mechanical performance
present study is to move on to the multi-objective of the PG mixtures on pavement conception.
(environmental and economic) and multi-scale In the second section, we present results obtained
(materials and structures) optimization of the use of PG from mixture design experiments, the mathematical
as a pavement material. This work will focus on the use modeling of desired responses as well as domains
of the PG produced at Jorf Lasfar plant. Indeed, the use meeting the required criteria for using the mixture
of this material could provide interesting solutions to material either as road subgrade layer or foundation
the decrease of availability of primary materials within layer. We also discuss the effects of soil bearing

1 Currently known as the Ministry of Equipment and Water.

118 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

capacity, traffic level and mechanical properties of PG from Sonasid, a steel industry, adjacent to OCP’s Jorf
based materials on pavement structures’ design. Lasfar industrial complex. Its water content is about
We then draw some conclusions and present some 0.03%, so it is a dry material. Its impact resistance is
perspectives on work continuation. determined at LCC by [17]:
 Micro Deval Index : MDE (NF P 18-572) = 7
2. Materials and Methods
 Los Angeles Index : LA (NF P 18-573) = 16
2.1 Raw Materials They are therefore very hard aggregates.
GTR Class [17]: This is an industrial by-product, so
2.1.1 Jorf Lasfar PG
according to the road earthworks guide it is an F8 class
In Morocco, different qualities of PG are being
material.
produced at Safi and Jorf Lasfar industrial complexes.
2.1.4 Sand
The first experiment in Safi having given encouraging
It is a crushed siliceous sand, of a 5 mm maximum
results [13], the research study was extended to Jorf
diameter. Its fine element content (< 80 μm) is about
Lasfar the largest production site. PG taken from
9%, almost zero water content. Apparent density is
production unit of Jorf Lasfar has 30% water content.
about 1.60 g/cm3 [19].
The phosphogypsum consists of 95% dry weight of
calcium sulfate dihydrate determined by diffractometric 2.2 Characterization Methods
analysis [16]. Minerals of quartz, dicalcium phosphate,
For chemical characterization of both PG and steel
iron, magnesium and aluminum sulfates constitute the
slag, Plasma emission with optical detection (ICP OES)
minority solid fraction. The phosphogypsum is presented
is the method used to quantify sulfur trioxide SO3,
in the form of crystals tabular to parallelepipedal
calcium oxide CaO, phosphoric anhydride P2O5,
elongated crystals [16]. This is an industrial coproduct,
alumina Al2O3, potassium oxide K2O, silicon dioxide
so according to the road earthworks guide GTR [17], it
SiO2, magnesium oxide MgO, iron oxide Fe2O3, and
is an F5 class material.
sodium dioxide Na2O by continuous flow.
2.1.2 Cement
Whereas pH measurement has been done after 5 min
It is a composite Portland cement rated CEM II,
agitating of 10 g dry sample in 100 mL distilled water,
coming from Lafarge-Holcim Bouskoura’s plant near
then using direct pH-meter.
Casablanca (Morocco). The chemical characteristics are
listed in Table 1. Its loss on ignition at 1,000 °C is 3.0%, 2.3 Raw Materials Properties
insoluble residue 14.2%, Blaine specific surface Table 1 shows some of raw materials’ chemical
3,908.7 cm²/g, mixing water 30.8%, expansion 1.2 mm, composition and their pH.
start of setting time 229.7 min, end of setting time 330.9
Table 1 Raw materials characterization.
min. In terms of mechanical resistance, simple
PG C SS
compression strength (measured according to NF EN
pH 2.85 11.22
13286-41) at 2 days is 14.3 MPa, 33.2 MPa at 7 days
SO3 (g/100g) 39 2.8 0.26
and 41 MPa at 28 days.
CaO (g/100g) 26 47.7 26
2.1.3 Steel Slag Al2O3 (g/100g) 0.16 8.52
In metallurgy, especially steel, slag is a solid by- K2O (g/100g) 0.06 1.2 0.16
product of smelting, refining, treating or shaping metals SiO2 (g/100g) 0.02 0.52
at high temperature. It is a various oxides mixture that MgO (g/100g) 0.02 2.2 4.73
floats on the molten metal or is detached from it when Fe2O3 (g/100g) 0.01 24
Na2O (g/100g) 0.4
it is processed at high temperature. Our steel slag comes
 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 119

2.4 Mechanical Criteria for Use as Road Materials  Cs > 1.2 MPa to take into account average site
traffic,
The objective is to evaluate the effect of the variation
 (Ts360d, E360d) at least in zone T1 of the
of the proportions of the constituents of the various
classification chart (Fig. 2).
mixtures on the physical and mechanical characteristics
that allow mechanical dimensioning and stability 2.5 Material Optimization Procedure
monitoring, as defined by roads design guides. The
criteria for subgrade layers and foundation layers are 2.5.1 Bearing Capacity of PG
presented hereafter. We have conducted load-bearing tests on raw PG
2.4.1 Mechanical Criteria for Subgrade Layer (according to NF P 94-078) the results of which, shown
Mixture criteria for subgrade layer use are [19]: in Table 2, proved that raw PG does not meet
 CBR-4d > 20 and CBR-4d/IBI > 1, mechanical criteria of road use from GTS 2000 [19]
 Cs28d > 1MPa, and CFTR 2007 [20], as reported in the literature.
 (Ts360d, E360d) at least in zone 5 of the For this reason, a cement treatment is necessary, for
classification chart (Fig. 1). the purpose of stabilization and required resistance
2.4.2 Mechanical Criteria for Pavement Layer achievement. This cement addition also neutralized the
Mixture criteria for foundation layer use are similar acidity due to the PG. Moreover, a granular correction
to above, except [20]: using conventional materials (sand, aggregates) is also

Fig. 1 Classification chart of treated materials for subgrade layer.

120 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

Fig. 2 Classification chart of treated materials for foundation layer.

Table 2 Bearing tests results of on raw PG.  Determine key factors in a new product design.
IPI CBR-4d CBR 4d/IPI 𝐆𝐯 (%) It is therefore an effective tool in the case of this
PG 68 46 0.67 0.10 study where experimentation is associated with a
recommended to achieve a better granular skeleton and modeling objective. We are dealing with a 3-
thus higher density. In order to remain in a circular component mixture which sum of proportions is equal
economy approach, other available by-products steel to 100%, therefore the mixture design under constraints
slag (SS), washing tailings (T) have been considered. [21] turns out to be the best design suited for this study.
In this article, our choice went to steel slag (SS) and 2.5.3 Responses
sand (S) for the 1st and 2nd formulation respectively. The responses targeted throughout this study are
2.5.2 DOE Method those defined in section 2.4:
In order to evaluate the variation effect in  IBI: Immediate bearing index according to NF P
constituents’ proportions on the responses, and to 94-078 (Fig. 3),
derive mathematical models of the responses, we have  CBR-4d: Californian bearing ratio after 4 days of
adopted DOE method [21], which is a statistical water immersion according to NF P 94-078 (Fig. 4),
approach offering several tools making it possible to  Gv: The vertical swelling after 4 days of water
control a multi-parameter problem, while following an immersion with overload (%) according to NF P 94-
experimental program. 078 (Fig. 4),
It aims to:  Cs7d, 28d: Simple compressive strength at 7 and
 Minimize experimental measurements’ number 28 days (MPa) according to NF EN 13286-41 (Fig. 5),
compared to a classic approach, while ensuring a  Ts28d: Diametral tensile strength at 28 days (MPa)
quality of results, if not better, at least identical, according to NF EN 13286-41 (Fig. 6),
 Put the focus on expected results, before  E28d: Elastic modulus at 28 days (MPa) according
experimenting them, to NF P 98-232-3 (Fig. 6).
 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 121

Fig. 6 Diametral tensile strength and elastic modulus (NF

P 98-232-3).

Fig. 3 Immediate bearing index (NF P 94-078). 2.5.4 Factors and Fields of Study
Factors considered throughout this research are the
principal components necessary for the mixture
formulation: phosphogypsum, steel slags or sand,
cement, and more precisely their content in mixtures.
Choosing ranges of study for all factors was
determined in such a way as to maximize the use of PG
and minimize the cement use:
 45% < PG < 70%,
 4% < C < 7%,
 23% < SS or S < 48%.
The optimal matrix of experiments (mathematically)
Fig. 4 Californian bearing ratio after 4 days of water is defined to achieve reliable and efficient results, while
immersion and vertical swelling after 4 days of water
immersion with overload (NF P 94-078).
minimizing the number of tests [21]. Our experimental
matrix is represented in Table 3, and field of study on
the simplex in Fig. 7.
The granular envelope of the 9 experimental points
in Fig. 8 turns out to be continuous and well distributed.

Table 3 Table representing experimental points.

Formulation PG (%) SS or S (%) C (%)
1 70.000 26.000 4.00
2 51.625 42.125 6.25
3 58.250 36.250 5.50
4 64.125 29.625 6.25
5 70.000 23.000 7.00
6 53.125 42.125 4.75
7 48.000 48.000 4.00
8 45.000 48.000 7.00
9 64.125 31.125 4.75
Fig. 5 Simple compressive strength (NF EN 13286-41).
122 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

therefore necessary to determine optimal Proctor

density of experimental mixtures before studying
targeted responses.
Proctor test requires density determination of at least
five points at different water contents. Performing these
tests for each of the nine experimental mixtures
requires 45 experimental points. In order to save time
and gain in productivity, we opted for a mathematical
modeling, which would make it possible to obtain
Proctor results based on DOE method. We worked on
the mixture (PG-S-C). The modeling consists of taking
a sample of 15 points with different water contents to
build a mathematical model which predicts Proctor
densities and Proctor water content values.
Fig. 7 Graphic representation of the 9 experimental points
The chosen model should establish an equation
in the study area. considering PG, sand and cement dosages as factors, in
addition to water content to give the mixture dry
2.6 Sample Preparation and Water Content
density. Then, this equation is used to obtain optimal
Simple compression test Cs is carried out for a dry density and Proctor water content for each mixture
material density set at 98.5% of Proctor dry density (NF processed.
P 98-230-1), while tensile strength test Ts is set at 96% Targeted results, Proctor optimum dry density and
of Proctor dry density (NF P 98-230-3) (Fig. 9). It is water content, are shown in the next section.

Fig. 8 Particle size distribution of PG, SS, and the 9 mixtures of the PG-SS-C formulation.
 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 123

= 120 MPa and µ = 0.35).

 As for the traffic level, designated by Tr, the
values of the annual average daily traffic TMJA, the
class T, and the number of axles NPL of 13 tons of
heavy goods vehicles soliciting the roadway refer to the
values from the middle of the traffic class T2 and T3 of
PNM 2019 [22] Tr-: TMJA = 65; T3; NPL = 280.000
and Tr+: TMJA = 180; T2; NPL = 790.000.
 For treated phosphogypsum performances, the two
values of Elastic modulus E and tensile strength Ts at
360 days refer to the high and low level of PG treatment
by varying cement dosage (PG-: E = 2,000 MPa; Ts =
0.2 MPa and PG+: E = 8,000 MPa; Ts = 0.35 MPa).

Fig. 9 Moulding-demoulding of simple compressive

3. Results and Discussion
strength test tubes.
3.1 Proctor Model Optimization
2.7 Pavement Structure Optimization
The actual Proctor test points are presented in Table
In order to determine the optimal type of pavement 4. For each experimental point, Proctor optimum dry
structure allowing better road use of phosphogypsum, density has been measured2.
a parametric study was carried out on 3 types of The quadratic model is judged to be the most suitable
pavement structure (Fig. 10), varying three factors: for our case study since it has a larger adjusted
bearing capacity of the soil (BC), traffic level (Tr), and correlation coefficient [21]. Moreover, its p-value [21]
mechanical performance of phosphogypsum mixes (PG) is less than 0.05, which proves that this model is
[14]. statistically valid. It has an adjusted R² coefficient [21]
 For the supporting soil designated by BC, the two of 88%. A verification of the model for our
values of the Elastic modulus E and the Poisson’s ratio experimental points shows that the difference between
µ refer to the library adopted by the ALIZE-LCPC actual results and density obtained through the model
software [15] (BC-: E = 50 MPa; µ = 0.35 and BC+: E does not exceed 0.01 t/m3 on average.

Fig. 10 Types of pavement structures considered in the parametric study [22].

2 Complete data can be requested from the corresponding author.

124 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

Table 4 Experimental points of Proctor test modeling. mean values. Cement effect can be considered as
Point PG S C w negligible compared to those of other factors.
1 0.7 0.26 0.04 0.064 The coefficients a and b obtained are low, which
2 0.7 0.26 0.04 0.085 means that dry density curves as a water content
3 0.51625 0.42125 0.0625 0.08
function are relatively flat nearby Proctor optimum.
4 0.5825 0.3625 0.055 0.086
This is a material positive characteristic which allows
5 0.5825 0.3625 0.055 0.115
an easy implementation on site.
6 0.64125 0.29625 0.0625 0.12
7 0.7 0.23 0.07 0.16
The cement treatment allows a significant reduction
8 0.53125 0.42125 0.0475 0.127 in the water content of the mixture. Sand has a positive
9 0.53125 0.42125 0.0475 0.14 influence on increasing dry density. The mixture optimum
10 0.48 0.48 0.04 0.09 water content is relatively low, which contributes to
11 0.48 0.48 0.04 0.102 durability improvement of pavement layers.
12 0.45 0.48 0.07 0.08 Thus, the results of this Proctor mathematical
13 0.45 0.48 0.07 0.07
modeling led to water optimum values wopm ranging
14 0.64125 0.31125 0.0475 0.06
between 8.44% and 9.55% whereas associated optimal
15 0.64125 0.31125 0.0475 0.1
Proctor densities Dopm ranged between 1.61 t/m and
Based on regression equation, and fixing the other 1.82 t/m .
parameters (PG-S-C) for each mixture, we go back to a
3.2 DOE Results
second order polynomial equation we can solve to
determine the couple (wopm, Dopm): 3.2.1 Bearing and Swelling Test
D(w) = aw2 + bw + cte (2) For each of the studied formulations, we carried out
Wopm = -b/2a; Dopm = D(wopm) (3) an immediate bearing test IBI, a CBR after 4 days of
The main effects diagram (Fig. 11) defines each water immersion and measured vertical swelling Gv.
factor (PG, S and C) and each process variable (w) main PG-SS-C and PG-S-C mixtures’ results are shown in
effect on density values: to have a high Proctor density Table 5. CBR-4d data can be requested from the author,
it is practical to seek a compromise between PG and range between 100 and 278 for PG-SS-C and
average values, S maximum values and water content between 115 and 322 for PG-S-C.

Fig. 11 Main effects diagram, Proctor modeling.

 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 125

Table 5 IBI and Gv for PG-SS-C and PG-S-C formulations. S-C mixtures at 28 days are presented in Table 7,
Mixture PG-SS-C PG-S-C whereas tensile strength at 28 days can be requested
Response IBI Gv (%) IBI Gv (%)
from the author.
1 80 0.7 160 0.1
2 138 0.4 191 0.3
3.3 Mathematical Response Models
3 121 0.2 124 0.1
4 111 0.6 145 0.4
5 102 0.8 138 0.5
Now that all experimental results are available, we
6 119 0.2 133 0.1 proceed by DOE method [21], to mathematically model
7 128 0.09 158 0.03 each of the targeted responses, continuously over the
8 140 0.6 149 0.2
entire study area.
9 98 0.4 101 0.2
Thus, responses of mathematical models for PG-SS-C
The addition of cement and steel slag considerably mixture are shown in Table 8, and for PG-S-C in Table 9.
improves the immediate bearing capacity, as well as
3.4 Model Verification Example: E28d
after 4 days of water immersion.
For mixtures with the same cement dosage (i.e., E28d response model is drawn in Fig. 12. Module
mixtures 1 and 8) the IBI goes from 102 to 140 and the increases with cement content, and at a fixed one, it
CBR-4d from 159 to 278, which shows steel slag has
Table 6 Experimental results of the simple compressive
influence on bearing capacity, which is clearly strength of the PG-SS-C and PG-S-C mixtures at 7 days.
improved after 4 days of water immersion. The CBR- Mixture PG-SS-C PG-S-C
4d/IBI ratios are all greater than 1, which is a criterion Response
Cs 7d Cs 7d (bis) Cs 7d Cs 7d (bis)
(MPa) (MPa) (MPa) (MPa)
for the treatment durability [18].
1 0.18 0.16 0.22 0.20
The vertical swelling Gv varies between 0.094% and 2 0.27 0.26 0.29 0.32
0.84%, which represents acceptable values for subgrade 3 0.24 0.22 0.44 0.45
layer use. Indeed, the GTS 2000 [20] and CFTR 2007 4 0.27 0.27 0.37 0.37
[18] guides tolerate a maximum swelling of 5%. 5 0.21 0.21 0.86 0.96
6 0.22 0.21 0.48 0.62
The results of the PG-S-C shown in Table 5, prove
7 0.21 0.215 0.55 0.53
that a high dosage of sand gives greater bearing indexes
8 0.35 0.35 0.79 0.89
in addition to increased dry densities. However, sand 9 0.2 0.19 0.40 0.37
alone does not have a dominant effect.
Table 7 Experimental results of tensile strength and
3.2.2 Compressive Strength stiffness modulus for PG-SS-C mixture at 28 d.
The simple compressive strength values at 7 days for Mixture PG-SS-C PG-S-C
PG-SS-C and PG-S-C mixtures, carried out twice, are E 28d E 28d (bis) E 28d E 28d (bis)
Response
shown in Table 6. (MPa) (MPa) (MPa) (MPa)
1 1,263.6 1,300 1,098 1,053
Based on these results, we conclude that at the age of
2 1,048.4 1,100 1,349 1,166
7 d, all mixtures’ simple compressive strength does not 3 1,046.5 1,050 1,861 1,812
exceed 0.4 MPa, which means that time for circulation 4 919.1 950 708 726
exceeds 7 d. According to GTS 2000 [20] and CFTR 5 785.8 801 1,376 1,359
2007 [18], the simple compressive strength should 6 806.6 810 600 590
exceed 1 MPa at the age of circulation. 7 1,358.5 1,415 530 535
3.2.3 Elastic Modulus 8 1,302.6 1,298 1,155 1,042

Elastic modulus E28d results for PG-SS-C and PG- 9 910.6 899 647 601
126 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

Table 8 Responses mathematical models for PG-SS-C mixture.

Adjusted p-
Mathematical models of the responses of the PG-SS-C mixture (the dosages are in %)
R² value
IBI = 1.14 × PG – 14 × SS + 4,324 × C + 0.08 × PG × SS – 47,971 × PG × C – 48,169 × SS × C 97% 0.002
Gv = 0.056 × PG + 0.0039 × SS + 11.9 × C – 0.13 × PG × C - 0.12 × SS × C 97% 0.017
Cs7d = 0.015 × PG - 0.051 × SS + 20.8 × C + 0.0032 × PG × SS - 0.23 × PG × C - 0.23 × SS × C + 0.00013 × PG
96% 0
× SS × C
E28d = 150 × PG – 598 × SS + 193,016 × C + 33 × PG × SS - 2,152 × PG × C - 2,146 × SS × C 92% 0

Table 9 Responses mathematical models for PG-S-C mixture.

Mathematical models of the responses of the PG-S-C mixture (the dosages are in %) Adjusted R² p-value
Gv (%) = 1.06 × PG + 18.57 × S - 3,222.45 × C – 0.79 × PG × S + 35.44 × PG × C + 34 × 67 × S × C
99% 0
+ 0.034 × PG × S × C
IBI (%) = 22.75 × PG - 125.31 × S + 37,753.13 × C + 6.72 × PG × S - 420.44 × PG × C – 420.26 × S × C 99% 0,01
Cs7d (MPa) = -0.92 × PG - 13.45 × S + 2,271.59 × C + 0.57 × PG × S - 24.94 × PG × C - 24.36 × SC × C
95% 0
- 0.0267 × PG × S × C
E28d (MPa) = -7,645.4 × PG - 136,430 × S + 23,774,900 × C + 5,851 × PG × S - 261,479 × PG × C
97% 0
- 255,924.2 × S × C - 252.25 × PG × S × C

Fig. 12 Elastic modulus E28d modeling for PG-SS-C mixtures.

increases with steel slag content, and decreases with PG We notice that relative deviation (less than 8%) remains
content. acceptable (Table 10). 360 days value is the sizing
We conducted an experimental verification for a value, since it will be reported on the classification
special point, chosen inside the experimental domain, chart mentioned above (Fig. 12).
by making and testing new samples at 28 d, 90 d and 𝐸
= 0.65 (4)
360 d to determine the experimental values and 𝐸
compare them with model values, which were 𝐸
= 0.7 (5)
extrapolated from 28 d, following Eqts. (4) and (5) [20]. 𝐸
 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 127

Table 10 Values verification of the mathematical model mentioned above, that can be combined, if needed, with
E28d for a specific mixture PG-SS-C.
economic, environmental and radiological criteria.
Model value Experimental Relative
(MPa) value (MPa) deviation
3.6 Pavement Structure Optimization
E28d 1,184 1,125 5%
E90d 1,275* 1,383 7.8% 3.6.1 Factors Effect
E360d 1,821** 1,975 7.8% For the three types of pavement structures (Fig. 10),
** Extrapolated from E28d following Eq. (4).
soil bearing capacity has an important effect on
* Extrapolated from E360d following Eq. (5).
pavement total thickness (Fig. 14), in particular that of
3.5 Mixtures Verifying Mechanical Criteria the treated phosphogypsum layer, which decreases
when going from a low level of soil bearing capacity
Having modeled all needed responses, we can
( 𝐵𝐶 ) to a high level ( 𝐵𝐶 ). Mixed-structure
graphically determine domains where materials’
pavements are generally thinner than other types of
properties verify the criteria defined in section 0.
pavements, in all configurations. As hydraulic
Mixtures for PG-SS-C formulations verifying criteria
binder treated sub-base pavements have two layers of
either for subgrade or foundation layer are represented
treated phosphogypsum, they consume more of this by-
in Fig. 13. It shows that a wide area respects all criteria
product. Indeed, this type of pavement can consume
up to twice more phosphogypsum than pavements
with a mixed structure, and up to 2.3 times more
phosphogypsum than pavements with an inverse
structure.
For the three types of pavements, the total thickness
of the pavement and that of the treated phosphogypsum
layer increases with heavy traffic. In this particular
case, inverse pavement structure can be more
interesting (Fig. 14).
In general, the total thickness of the pavement and
that of the treated phosphogypsum for mixed structures
tend to decrease by increasing the rigidity of the
phosphogypsum mixture.
The total thickness of inverse pavement structure
increases by increasing the rigidity of the PG based
material, but it is not the case of pavements with sub-
base treated with hydraulic binder. For a softer
phosphogypsum mix, inverse pavement structure tends
to be thinner than the one with hydraulic binder treated
sub-base, for a more rigid phosphogypsum mix, the
first is thicker than the second.
We can gain up to 15 cm in total thickness by
adopting a mixed structure compared to an inverse one,
Fig. 13 PG-SS-C: suitable domain either for subgrade or and this difference is particularly noticeable when the
foundation layer. phosphogypsum mixture is more rigid.
128 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

70
BC- and Tr- BC+ and Tr- BC- and Tr+ BC+ and Tr+

Mixed structure 60

50
Thickness in cm

Inverse structure 40

30
Sub-base treated
with hydraulic 20
binder
10
PG thickness
0
PG- PG+ PG- PG+ PG- PG+ PG- PG+
Fig. 14 Effect of each factor on PG layer thickness and total pavement thickness.

140

120

100
Relative price in%

80

60

40

20

0
BC- Tr- PG- BC+ Tr- PG- BC- Tr+ PG- BC+ Tr+ PG- BC- Tr- PG+ BC+ Tr- PG+ BC- Tr+ PG+ BC+ Tr+ PG+

Mixed Structure Inverse Structure Structure with subbase treated with hydraulic binder
Fig. 15 Relative price estimation of the mixed/inverse/treated pavement structure compared to a conventional structure (thick
bituminous).

There is also a gain up to 9 cm in total thickness by 3.6.2 Economic Gain

adopting a mixed-structure pavement compared to a We have carried out, on the basis of the Moroccan
pavement with a sub-base treated with hydraulic binder market prices, a basic cost analysis, based on the
(Fig. 14). following hypothesizes:
The substitution by PG layers allows preserving  Price calculation is relative to the most used
conventional resources (gravels, sand). Actually, PG conventional structure in the Moroccan context, which
layer in a mixed structure can constitute from 53% to is the thick bituminous [22], considered equal to 100%;
60% of total pavement thickness, whereas in inverse  The price is that of materials implemented on site
structure, it represents from 40% to 54%, and in a excluding transport costs;
treated PG sub-base from 87% to 89%.  We assume a local availability of materials;
 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization 129

 We consider the price of PG as zero; treated phosphogypsum mixtures on pavement design

 We take Ref. [23] as the basis of prices on the for three different pavement structures (mixed, reverse
Moroccan market. and structure with treated sub-base) and determined
We note that pavements with a sub-base treated with that the best to adopt for maximizing PG valorization,
hydraulic binder have an undeniable cost advantage and thus saving primary resources, is the pavement
(Fig. 15) compared to other types of structures. Indeed, structure with subbase treated with hydraulic binder.
this type of pavement can be up to 52% less expensive We will continue this work using phosphate tailings
than mixed pavements, and up to 43% less expensive as granular corrector and RHB instead of cement as a
than reverse structure pavements. Mixed structures are binder. We will also carry out a complete
the most expensive variant, which was expected given environmental and health assessment of the different
the cost of the bituminous layer which represents at mixtures in order to validate their environmental
least 40% of the thickness of the pavement. harmlessness and the safety of their use.
3.6.3 Parametric Study Conclusion We will then proceed with the construction of new
Finally, the parametric study made it possible to pilots at different locations, having different soil
conclude that whatever traffic, soil bearing capacity bearing capacities, on classified national road network,
and the treated PG performances, pavements with using the three types of pavement structures evaluated
subbase treated with hydraulic binder are very in this paper.
competitive compared to pavements with reverse and
Acknowledgments
mixed structures.
This work is funded by OCP group as part of pilot
4. Conclusions and Perspectives projects for phosphogypsum valorization in road
In this article we studied the possibility of using construction. It also benefited from a technical support
materials consisting of Jorf Lasfar’s PG, mixed with a from the Roads Department, in particular the National
granular corrector (either sand or steel slag) and treated Center for Road Studies CNER, which we warmly
with cement, as a material for road construction (either thank, and a grant from the national scientific and
as a subgrade or foundation layer). technical research center CNRST. We do also want to
We conducted the experimental study using the DOE warmly thank the GTR, partner in the project, for help
method, which optimizes the total number of trials and guidance from its laboratory LCC for the first
while allowing obtaining reliable mathematical models experimental campaign.
for the studied responses.
References
This made it possible to determine models for the
[1] Rutherford P. Dudas M. and Samek R. 1994.
physical and mechanical responses used in the design
“Environmental impacts of phosphogypsum”. The Science
of pavement materials. We verified the models’ quality of the Total Environment. pp. 149 (1-38).
by testing a new mix not used in the DOE experimental [2] Ferguson. F. 1998. “Phosphogypsum - an overview.”
points. Comparison between experimental values and Proceedings of the Second International Symposium on
Phoisphogypsum. pp. 117-130.
predicted ones showed less than 8% relative deviation. [3] May, A., and Sweeny, J. 1984. “Assessment of
Compositions’ domains that meet the criteria (Fig. 13) Environmental Impacts Associated with Phosphogypsum
appear to be large enough to allow further design in Florida.” In The chemistry and Technology of Gypsum,
edited by K. Ra. Philadelphia: ASTM Special Technical
criteria (environmental, cost, …).
Publication, pp. 116-39.
Furthermore, we evaluated the effects of traffic level, [4] Senes. 1987. An Analysis of the Major Environmental and
soil bearing capacity and mechanical performance of Health Concerns of Phosphogypsum Tailings in Canada
130 Moroccan Phosphogypsum Use in Road Engineering: Materials and Structure Optimization

and Methods for Their Reduction. Ontario, Canada: Min. National de la Route, Dakhla, Maroc. (in French)
Environ. [14] Diouri, C., Echehbani, I., Lahlou, K., El Omari, K., and
[5] Vick, S. 1977. “Rehabilitation of a Gypsum Tailings Alaoui, A. 2022. “Valorization of Moroccan
Embankment.” In Proceedings of the Conference on the Phosphogypsum in Road Engineering: Parametric Study.”
Geotechnical Disposal of Solid Waste Materials. Journal Materials Today: Proceedings 58 (4): 1054-8.
[6] IFA. 2016. Phosphogypsum: Sustainable Management https://doi.org/10.1016/j.matpr.2022.01.084.
and Use (1st ed.). Paris: AE “Johnny” Johnston General [15] IFSTTAR. ALIZE-LCPC. 2003. Le logiciel pour le
Editor. dimensionnement des structures de chaussée.
[7] Office for Official Publications of the European http://www.alize-lcpc.com/fr/index.php. (in French)
Communitie. 2002. Design Guide—Road Construction [16] Moutawakil, A., Pineau, J., and Lahlou, K. 2003. “La
Based on Phosphogypsum and Fly Ash, Disposal recherche d’un procédé viable de valorisation d’un
Management System for Utilization of Industrial phosphogypse provenant de l’industrie phosphatière
Phosphogypsum and Fly Ash. EU, Project No.: marocaine.” Environnement, Ingénierie & Développement
LIFE98ENV/FIN/000566, Luxembourg. 129: 31-5. (in French)
[8] Havanagi, V., Silha, A., and Parvathi, G. 2018. [17] Laboratoire central des ponts et chaussées (LCPC).
“Characterization of Phosphogypsum Waste for Road Service d’études sur les transports, les routes et leurs
Construction.” Presented at Indian Geotechnical aménagements SETRA. 1992. « Réalisation des remblais
Conference IGC. http://www.igs.org.in:8080/portal/igc- et des couches de forme (GTR) - Fascicule 1-Principes
proceedings/igc-2018-bengaluru-proceedings/Theme09/T généraux » vol. 1. Paris: CEREMA (ex SETRA). (in French)
H-09-009.pdf. [18] Laboratoire central des ponts et chaussées LCPC, Service
[9] Peng, B., Yang, Z. C., Yang, Z. W., and Peng, J. W. 2020. d’études sur les transports, les routes et leurs
“Effects of pH and Fineness of Phosphogypsum on aménagements SETRA. 2007. « Traitement des sols à la
Mechanical Performance of Cement-Phosphogypsum- chaux et/ou aux liants hydrauliques : application à la
Stabilized Soil and Classification for Road-Used réalisation des assises de chaussées » (in French)
Phosphogypsum.” Coatings 10 (11): 1021. [19] Moussaab, M., Aitelcaid, H., Diouri, C., and Lahlou, K.
https://doi.org/10.3390/coatings10111021. 2021. Valorisation des Phosphogypses Marocains en
[10] Amrani, M., Taha, Y., Kchikach, A., Benzaazoua, M., and Technique Routière. PFE EHTP. (in French)
Hakkou, R. 2020. “Phosphogypsum Recycling: New [20] Laboratoire central des ponts et chaussées (LCPC).
Horizons for a More Sustainable Road Materials Service d’études sur les transports, les routes et leurs
Application.” Journal of Building Engineering 30: 101267. aménagements SETRA. 2000. Traitement des sols à la
https://doi.org/10.1016/j.jobe.2020.101267. chaux et/ou aux liants hydrauliques: application à la
[11] Messkini, S., Samdi, A., and Ejjaouni, H. 2021. réalisation des remblais et des couches de forme. (in
“Valorization of Phosphogypsum as a Road Material: French)
Stabilizing Effect of Fly Ash and Lime Additives on [21] Goupy J. 2017. “Introduction aux plans d’expériences.”
Strength and Durability.” Journal of Cleaner Production Collection: Techniques et ingénierie (5th ed.). Dunod.
323: 129161. https://doi.org/10.1016/j.jclepro.2021.129161. EAN: 9782100778027. (in French)
[12] Sfar Felfoul, H., Clastres, P., Ben Ouezdou, M., and [22] IMANOR. 2019. Projet de norme marocaine PNM
Carles-Guiberges, A. 2002. “Amélioration des 13.1.219. Dimensionnement structurel des chaussées
caractéristiques du phosphogypse en vue de son utilisation routières: Application aux chaussées neuves. (in French)
en technique routière.” Déchets, Sciences et Techniques 28: [23] Diouri, C., El Khoutayri, K., Lahlou, K., El Omari, K., and
21-5. (in French) Alaoui, A. 2021. “Valorisation de Sous-Produits
[13] Diouri, C., Lahlou, K., Alaoui, A., and El Omari, K. 2022. Industriels Marocains En Technique Routière.” Academic
“Valorisation du phosphogypse dans la construciotn Journal of Civil Engineering 39 (1): 104-11.
routière au Maroc.” In Proceedings of the 11ème Congrès https://doi.org/10.26168/ajce.39.1.24.