1. Introduction
In the agro-food industry, olive oil production is one of the industrial practices in Mediterranean countries that contribute to the economy at higher levels [
1]. During the extraction process, olive mill wastewater (OMW) is generated in higher amounts (in the successive steps of leaf removal, olive washing, grinding, beating and separation of oil) and considered as an environmental hazardous stream due to the high pollutant load, high concentrations of recalcitrant compounds, long-chain fatty acids, phenolic compounds, solids and biotoxicity [
2]. Consequently, OMW treatment is not an easy task, and nowadays, most of the waste streams are, on the one hand, considered to be byproducts with a potential for revalorization in the framework of a circular economy approach, whilst, on the other hand, the revalorization of OMW by adopting low-cost treatments appears to be nearly impossible due to its characteristics [
3].
Removal of solids from OMW is the first mandatory step, suggested to be performed initially, and possible by using coagulant materials or by centrifugation methods. Coagulation is defined as a chemical process used in wastewater treatment to remove suspended solids and to improve the removal of the chemical oxygen demand (COD). By employing this process, the tendency of small particles in an aqueous suspension to combine with one another can be significantly improved. The interaction among the particles results in the formation of larger aggregates that can be removed from the water by sedimentation and/or filtration as the process of coagulant cumulation to wastewater, in order to destabilize the colloidal particles promoting agglomeration, forming larger particles, which can be easily removed by sedimentation [
4,
5].
Several synthetic (inorganic and synthetic organic) and natural organic coagulants (NOC) are commonly used in conventional water treatment processes; however, synthetic inorganic coagulants such as FeCl
3, polyaluminum chloride (PAC) and aluminum sulfate, Al
2(SO
4)
3·14H
2O (alum), are the most accepted and practiced in large-scale wastewater treatment plants [
6]. Alum is efficient in turbidity removal [
7] which will enhance the photocatalysis process. Azbar et al. [
8] treated OMW with the coagulation and flocculation method prior to anaerobic degradation using alum and achieved 47% removal of COD, 27% total phenols, and 39% total organic carbon (TOC). Chiavola et al. [
9] worked on OMW with several coagulants, and they observed that alum efficiency was less when compared to acid crackling; however, the raw OMW in their work contained high COD, at 69.4 g/L, and they achieved 20% of COD with 400 mg/L of the alum dosage and a pH of 8. Vuppala et al. [
5] achieved 57.1% of COD removal with an 800 mg/L dosage of alum at pH 4.5. This shows that OMW requires extensive optimization since the characteristic changes depending on the month, olive fruit type, extraction method and region. Considering these facts, further optimization of coagulation and flocculation of OMW is needed since the required dosage of coagulant, optimal pH and other suitable conditions should be optimized [
5]. In fact, this process consumes a lot of time and resources, leading to unsustainable operating cost issues in factories. This prevents the successful use of the coagulation and flocculation process since uncertainty prevails. Statistical techniques should be employed to study the optimization problem in a feasible way, but stochastic processes require very large sample data to be supported [
10,
11].
Considering, together, the need and difficulty in selecting a suitable coagulant dosage for OMW treatment, one of the most popular statistical techniques, that is, grey relational analysis, was selected in this work, targeting the optimization of controlled parameters such as TOC, COD, phenols, and turbidity. The grey system theory looks at each stochastic variable as a grey quantity that varies within a fixed region and within a certain time frame, and parallel to this, each stochastic process is a grey process. The engineering concept of a grey box is the basic concept of the grey system theory. Uncertain or not well-known relations among input and output parameters of the process refer to grey relations, or in other words, a relationship among the parameters of a grey system, where the proximity or closeness of grey data sequences represents different unbound parameters of the system. In this paradigm, greyness implies incompleteness [
12]. In this work, TOC, COD, phenols and turbidity are taken as grey parameters that affect the amount of coagulant dosage to be used.
In the engineering and technical literature, the most popular technique is Deng’s grey relational analysis (GRA) model, permitting a multi-objective optimization and evaluation of process parameters. Some applications of the GRA model are as follows: (i) optimization of powder metallurgy processing parameters of the Al
2O
3/Cu composite using the GRA model coupled to the Taguchi method [
13], in order to optimize electrical discharge machining process parameters; (ii) optimization of the process parameters for machining aluminum hybrid composites using electrical discharge machining (EDM) [
14]; other researchers [
15,
16] also led works on EDM using the GRA model coupled with the Taguchi method for the multi-optimization of the micromachining process parameters; (iii) principal component analysis combined with multi-objective optimization of process parameters for hard turning of AISI 52100 steel was carried out [
11]; (iv) Aslantas et al. [
17] applied micro-milling to TI-6Al-4V, whereas another researcher [
18] used the analytic hierarchy process (AHP) integrated GRA model to analyze the integrated cascade utilization system of waste geothermal water; (v) the relationships between some parameters to study the degree of the relative importance of the influencing factors about fractal dimensions for super pulverized coal particles were studied [
19]. The list is only a selection of case studies that involved the GRA model or variants to seek successful process optimization.
Indeed, the common point of all processes is the high difficulty to select appropriate parameters to satisfy all the conditions. In such cases, the uncertainty of a system and the incompleteness of information could be related using grey relational analysis. Discrete data and multi-variable input can be processed for uncertainty and can generate discrete sequences for the correlation analysis of such sequences [
20]. Therefore, it could be defined as a measurement method that discusses the consistency of an uncertain discrete sequence and its target.
The present work reports multi-response optimization using GRA in the pretreatment of real olive mill wastewater through the coagulation–flocculation process by alum. The coagulation–flocculation method showed high removal rates of the targeted pollutants (TOC, turbidity, COD and total phenols). The experiments were performed by varying the coagulant dosages gradually from a low (100 mg/L) to a high amount (2000 mg/L), and the selection of the optimum amount of coagulant dosage was carried out using grey relational analysis. There is no literature available, so GRA is introduced for optimization of the coagulation and flocculation process to treat OMW. It was confirmed that GRA analysis is a potential and innovative statistical method to identify the optimal operating conditions without any errors.
4. Conclusions
OMW was treated through the coagulation and flocculation process using alum as a coagulant, with the target to find an optimum coagulant dosage among the range from a low dosage (100 mg/L) to a high dosage value (2000 mg/L) at constant pH, in small batches. To validate the optimal operating conditions, grey relational analysis (GRA) was adopted.
The highest value of GRG was obtained for the coagulant dosage of 600 mg/L. In these conditions, reduction performances of the considered multiple output characteristics (TOC, COD, phenols and turbidity) appeared maximized. In numbers, the obtained removal rates of turbidity, COD, TOC and total phenols were 96.31%, 64.63%, 23.77% and 66.15%, respectively. The results obtained from the grey analysis were validated using the same operating conditions on a bigger batch. Concluding, GRA appears to be a proper and useful tool to optimize operating conditions, viz., coagulant dosage, agitation times and agitation speed, to those not well-known processes affected by multi-parameter responses, such as the coagulation and flocculation of olive mill wastewater, and can improve the reliability of the process to be used safely as a proper pretreatment step in situ in olive oil factories.