Application of lean Six Sigma to improve the dense medium separation performance at a diamond processing plant in Namibia

Abstract

Lean Six Sigma (LSS) is a rigorous, data-driven, and results-oriented process improvement philosophy. To date, there is no evidence of a study of the application of LSS in a diamond plant. One of the most common recovery methods in the diamond industry has been dense medium separation (DMS), also known as heavy medium separation (HMS). The main purpose of the study is to investigate how to implement LSS to improve the dense medium separation performance at a diamond plant. Further, we investigate the factors impacting DMS by finding the most relevant factor/s and applying the principles of Lean and Six Sigma to improve the performance of dense medium separation (DMS) at a diamond plant in Namibia. The research found out that factors affecting DMS include cyclone pressure, medium density, viscosity, and worn cyclone parts and maintaining these parameters within narrow limits in relation to the correct set points is beneficial. Also, the study established that, operational conditions (cyclone pressures and medium densities) are the topmost factors affecting DMS followed by viscosity and lastly, equipment dimensions. The application of LSS reduced downtime delays and boosted the associated production, thus achieving cost-efficiency, improved DMS performance and reducing waste. Diamond mines are found in large numbers in Africa, and this study will help the mines to improve their efficiency. Further, using the lens of theory of swift and even flow, the LSS is applied in the diamond company to streamline the DMS performance. Further, it poses key implications for future LSS research.

Keywords:

• Lean Six Sigma
• viscosity
• dense medium separation

Public Interest Statement

In most diamond mines, one of the most critical processes is dense medium separation. To separate the diamonds from the gangue in the run-of-mine (ROM) ore, a multi-stage procedure involving categorization and the dense medium separation must be carried out. A leading diamond mine in Namibia has been experiencing extreme fines and viscosity issues in the DMS (Dense medium separation) circuit. To treat the run-of-mine (ROM) ore, the diamond mine has set up a DMS station. It has been identified that Zone 5 is a high clayish ore body containing fatty pockets of clay. Due to the clay material, the ore causes chute blockages from the front end of the plant, clay builds up in the wet 100-ton bin. This results in viscosity and medium stability problems in the DMS section, which can cause inefficiency. This study uses Lean Six Sigma to improve the Dense Medium Separation performance at a diamond processing plant

1. Introduction

Lean Six Sigma (LSS) is a philosophy applied by organizations aiming at enhancing quality, managing variation and eradicating waste (Antony et al., 2020; Sony et al., 2020). It is a rigorous, data-driven, and results-oriented process improvement methodology. Lean Six Sigma is an amalgamation of quality management theory and a strategy that is focused on eliminating variance, testing for faults, and increasing the quality of both finished products as well as the processes that produce them (Alexander et al., 2019). Many sectors of the economy, including mining, have benefited from Lean Six Sigma techniques (Sony et al., 2019). LSS offers a set of tools that are applicable to the mining sector to improve efficiency of operations. It can also be used to transform the business by cultivating a culture of excellence in the ranks of the company’s executives (Patel & Patel, 2021). However, the application of LSS in the diamond mining industry has been scant (Kęsek et al., 2019). In most diamond mines, one of the most critical processes is dense medium separation. To separate the diamonds from the gangue in the run-of-mine (ROM) ore, a multi-stage procedure involving categorization and the dense medium separation must be carried out. A leading diamond mine in Namibia has been experiencing extreme fines and viscosity issues in the DMS circuit. In order to treat the run-of-mine (ROM) ore, the diamond mine has set up a DMS station. It has been identified that Zone 5 is a high clayish ore body containing fatty pockets of clay. Due to the clay material, the ore causes chute blockages from the front end of the plant, clay builds up in the wet 100-ton bin. This results in viscosity and medium stability problems in the DMS section, which can cause inefficiency. In this study we propose a perspective and paradigm of Theory of Swift and Even Flow (TSEF), wherein this study can be served as an instrument of change and bolster the realisation of value from the application of LSS in diamond mines (Devaraj et al., 2013). The TSEF suggests that streamlining the flow of material and information in the process leads to process productivity increases. Similarly, the productivity falls if the flow is not streamlined and speed of flow is less than the designed one (Schmenner & Swink, 1998). The critical challenges faced by the diamond mines in Namibia has warranted to investigate a research question how to improve the DMS Separation performance in the diamond mining company? In finding answers to this question a critical examination of operations management principles that inform the understanding of diamond mining operations and sub operations and process through the implementation of LSS. Specifically, this study provides a practical guideline for managers, on how to implement LSS specifically on dense medium separation which is a critical process in a diamond processing plant. Further, it also strengthens operation management literature by contributing to theory. First, this study extends research of application of LSS on diamond mines, by explicitly elucidating and ranking the factors affecting the DMS on the processing plant. Second, it depicts how results of qualitative analysis can be integrated with Six Sigma and Lean philosophy to arrive at a solution, thus contributing to the theory of LSS implementation framework on a diamond mine. Third, this study provides a starting point for LSS implementation in a DMS process. Fourth, there has been a call recently to suggest practical intervention-based studies in operations management (Oliva, 2019). For this to happen the problem has to be defined in context. This will help in intervention strategies, which will produce mechanisms to achieve the outcome (Denyer et al., 2008). The study adds to this by defining the problem in the context of a diamond company, the intervention strategies wherein solutions were found using a qualitative study, DMAIC and Lean tools, followed by mechanisms that were developed for implementation and the outcome was analysed with respect to percentage recovery of spiking tracers. The remainder of the paper is organised as follows: Section 2 reviews streams of literature in LSS, Section 3 is devoted to methodology, Section 4 is concentrated on the case study. Section 5 is devoted to discussion and Section 6 to conclusion, limitation, and scope for future research.

2. Literature review

2.1. Theory of Swift and Even Flow (TSEF)

The quality guru Deming professed that improving the process will lead to better quality and lower costs. Further, the barriers between the departments have to be removed and also proposed anticipating various problems in the production of goods and services (Schmenner, 2004). The TSEF theorists argue that microeconomic theory is useful in understanding how labour and capital inputs translate into productivity. However, these theories are less useful when it comes to understanding many aspects of factory floor operations. Take, for instance, problems such as bottlenecks, variability in quality, variability of demand, efficiency of processes and equipment’s and workforce organization. The five basic laws governed by TSEF are (i) The law of variability. (ii) the law of bottlenecks (iii) the law of scientific methods (iv) the law of quality (v) the law of factory focus (Devaraj et al., 2013). TSEF suggests that the more even the flow of materials through the process, the more productive the process is. Thus, in simple terms, we can state the productivity of the production system increases with the speed of flow and decreases if the speed decreases (Schmenner, 2004). In a diamond mine process, flow should be maintained and hence operational excellence techniques such as LSS can be a great tool to improve the same.

2.2. Lean Six Sigma in diamond mining and processing

When it comes to improving processes, Lean and Six Sigma are two separate approaches (Cima et al., 2011). LSS is widely used in four important sectors of manufacturing, health care, human resource, finance and education (Singh & Rathi, 2018). LSS is used as business strategy to increase process performance, develop leadership, customer satisfaction, speed and costs and bottom-line results by improving quality (Walter et al., 2021) . Diamond mining firms that have implemented Lean management rely on Kaizen, or continued improvement, day in day out (Wilson, 2010). They use techniques used by Toyota, such as 5S, visual management, Andon, Heijunka, TPM, SMED, and Plan—Do—Check- Act, to precisely perform their tasks. The standard deviation sigma, a statistician’s parameter, inspired Lean Six Sigma (Antony et al., 2017). The Six Sigma technique aids diamond mining firms in increasing output and lowering expenses (Kęsek et al., 2019). The ultimate goal is to provide clients with a commodity that exceeds their expectations by removing process abnormalities. In Lean Six Sigma, low quality costs are eliminated (Duarte et al., 2012). For the most part, this kind of data is analyzed with the use of SPC, MSA, and DOE (Sunder, 2013). Using DMAIC, Lean Six Sigma effectiveness and reliability grows rapidly. It is possible that Lean and Six Sigma can benefit from one another despite their differences (Yadav et al., 2021). The two techniques have comparable action algorithms. In enhancing the efficiency of current process, Define,Measure, Analyse, Improve and Control (DMAIC) is employed. In Lean Management, Deming’s PDCA (Plan, Do, Check and Act) cycle is used as the basis for this method. LSS is a two-step strategy to continuous development in manufacturing and service processes that focuses on eliminating waste and decreasing variability (Antony et al., 2017). Lean Six Sigma seeks to maximize added value for customers while minimizing waste and reducing resource usage. Lean Six Sigma is a long-term attempt to steadily minimize variations in processes using a specified strategy. These two approaches, when combined, result in continuous improvement, and serve as the foundation for good system management in any firm desiring to grow rapidly (Duarte et al., 2012). When Lean and Six Sigma work together, the results are better than when they work independently. A streamlining team can use a mixture of these strategies to increase the promptness and productivity of each process within firms, resulting in more profit, lower costs, and tighter collaboration. There was some implementation of the LSS framework (Citybabu & Yamini, 2022; Singh et al., 2021), however, none of them focussed on implementation of LSS in a mining sector. In another study there was integration of Lean tools with requirements of ISO 9001:2015 standard as an operationalisation and support tool for a Quality Management System (QMS). The study proposed a model for integrating Lean tools and requirements of ISO 9001:2015. Such a methodology allowed the QMS to be more practical and dynamic, reinforcing the creation of value for the organisation. This integration has resulted in benefits such as 1) improvement of problem solving 2) improvement in internal communication waste reduction and 3) increase in productivity. Similarly in another study a combined methodology LeanDMAIC based on Lean Tools and DMAIC was developed. It was aimed at helping organizations to solve their problems easily and accurately. It was further demonstrated through a case-study implementation on an organization in the sector of wood products (Ferreira et al., 2019).There were many studies on mining sector mainly concentrated on Lean in mining (Andi et al., 2009; Dunstan et al., 2006; Flynn & Vlok, 2015; Lööw, 2015; Rocha et al., 2020), and Six Sigma in mining (Gargate et al., 2018; Hadidi et al., 2017; Nie, 2016; VanHilst et al., 2005). Maunzagona and Telukdarie (2017) study was conducted in diamond mining using the Value Stream Map and simulation modelling. This study even though focused on diamond mining, employed DMAIC and Root Cause Analysis (RCA). They suggested the need for practical LSS implementation studies in the application of diamond companies to improve the core processes. Besides, there are scant studies in LSS in diamond mines (Krishnan et al., 2020; Maunzagona & Telukdarie, 2017). In diamond mines, dense medium separation is one of the most critical processes. The performance of DMS is affected by many factors.

2.3. Factors affecting the DMS

The figure below depicts the features that have an effect on the DMS process and performance pointers (Sripriya et al., 2001) they highlighted the following elements listed in order of importance in Figure 1.

Figure 1. Elements affecting DMS source: (Sripriya et al., 2001).

From the illustration on the factors affecting DMS above, the cyclone operating conditions: medium density, medium to ore ratio and cyclone inlet pressure are the most prominent of the factors. The second most affecting factor is medium solids percentage, medium stability and rheology (clay contamination, viscosity). Finally, feed characteristics: determined by DMS feed rate control and particle size is the least affecting factor. Depending on the composition of the medium, rheology and stability are critical in Dense Medium Cyclone (DMC) separation. Stability measures how inclined the medium is, whereas rheology measures how difficult it is for the medium to flow.

2.4. Medium to ore ratio

Prior to entering the DM cyclone, the ratio in which the ore and dense medium are combined is critical. Individual particle mobility during passage through the cyclone is critical for successful separation of heavy and light fractions. Diamonds will be lost to tailings due to crowding and the resulting particle interference. To avoid crowding, trial and error approaches have revealed that a ratio of around 7 parts medium to 1 part ore is required. The ore and dense medium are blended in a specifically constructed mixing box that, when properly operated, keeps the cyclone feed’s dense medium to ore ratio at the correct proportion.

2.5. Cyclone pressure

An inappropriate inlet pressure will have a negative impact on the DM cyclone’s efficiency. Due to the extreme low inlet pressure, a high amount of light minerals will report to concentrate or sinks. Pumps or gravity can both be used to feed DM cyclones. Some of the primary circuits at plants are fed by gravity. Personnel at plants are, however, heading toward smaller, more compact circuits, all of which are pump supplied. A gravity-fed system, on the other hand, relies on an appropriate “head” of mixed feed to supply the cyclone with the proper input pressure. In other words, the DM cyclone is fed directly from the mixing box in gravity fed systems, which must be situated at a proper height above the cyclone to produce the required pressure. The gravity-fed DMS plants’ design prevents “too high” intake pressure and maintains the pressure at a constant level. As a result, the goal should be to get the highest feasible inlet pressure because it provides the best separation.

2.6. Density

The density of the medium has a direct relationship with the effectiveness of separation. The amount of concentrate reporting to sinks decreases as the medium density increases, and vice versa. The best separation is achieved at a reasonably high medium density; however, if the density is too high, diamonds will be lost to tailings.

2.7. Wear

The physical condition of the DM cyclone itself obviously has a significant impact on separation efficiency. The material fed into the cyclone is extremely abrasive, resulting in a rapid rate of wear. The vortex finder and spigot are the most vulnerable to wear. Shortening or even holing of the tube due to wear of the vortex finder causes short-circuiting of the feed to the overflow lid. Diamonds will be lost in the tailings as a result of this. As the spigot wears down, the discharge diameter grows larger and larger. The efficiency of separation degrades in direct proportion to the spigot size. As a result, increasing amounts of light minerals find their way into the concentrate. Due to its resilience to abrasive wear, high chromium content cast iron is commonly utilized in the manufacturing of DM cyclones. Nonetheless, regular inspections are required to evaluate the level of component wear and to ensure that worn components are replaced in a timely manner.

2.8. Viscosity in dense medium separation

During DM cyclone separation, viscosity of dense medium is crucial. Low medium density is caused by high medium viscosity, which increases the number of light minerals in the concentrate. It occurs because of fine ore particles, such as fine sand and clay, contaminating the dense medium. The increasing amount of concentrate generated, as well as the darkening of the medium, which tends to turn brownish in colour, are indicators of this. The consistency of the medium is porridge-like, and the flow is sluggish. Since the initial patent in 1912, cyclones have been used in mineral processing to separate particles of different densities, with the usage of magnetite or ferrosilicon as a vital part of the process to achieve the desired effect. MULTOTEC dense medium cyclone has been successfully employed to enrich and enhance the overall process, leading to improved revenues to user-plants for coal washing but also hard rock beneficiation (particularly diamond, iron ore and platinum). To upgrade moderately coarse particles in the range of 50 to 0.5 mm in size, DMCs have already been utilized by the mining industry for over 50 years. High-capacity centrifugal separators, dense medium vessels and baths cannot effectively increase the separation of small particles utilizing static density-based separators. DMCs are a good value for the money, and they require little in the way of operator involvement. As a result, DMC circuits have become increasingly popular and are now found in a wide range of mining and coal processing facilities. Approximately 40% of all diamond plants in the United States use DMCs, resulting in an installed capacity of more than 65,000 tons per hour. Considering the importance of DMS in a diamond processing plant, this study focussed on implementing LSS on it for the purpose of performance enhancement.

3. Methodology

The case study research is preferred when we require to obtain in-depth knowledge about a particular phenomenon. It is also used when the study consists of large number of variables, and these should be studied to understand the context under consideration (Robert. K. Yin, 2005). It can be thought of as an empirical inquiry, which is directed towards an emerging phenomenon in a real-life context (Şimşek et al., 2022). A point of distinction here is that the boundary between the phenomenon of study and the real-life context are not clearly evident (R. K. Yin, 2011). In terms of technique, it can consist of both qualitative and quantitative techniques to obtain a rich mix of data. Our attempt here is to understand the implementation of LSS to improve the performance of DMS by understanding the various contributory factors, which can influence the DMS performance over time. In exploratory case studies, it is pertinent to examine the theoretical background. This is helpful because it will help to concentrate on specific aspects of the case (Eisenhardt, 1989). The purpose of our study is to answer the main research question “how to implement LSS to improve the dense medium separation performance at the diamond plant?”. The criterion of our study is to investigate the factors impacting DMS, finding the most relevant factor/s and applying the principles of Lean and Six Sigma to improve the performance. The study uses a single case approach, this is because the researcher is interested in exploring the phenomenon in a live organizational setting to preserve its richness and content. Another point to consider is the choice between holistic and embedded design. The authors choose holistic design in order to consider the diamond organization and understand how LSS can be implemented in it for its efficacy. The single unit of analysis in this case would be the DMS performance, and the context surrounding the same is the diamond plant.

3.1. Data collection

There are various methods of data collection in case studies: 1) data collection instruments, 2) documentation, 3) archival records, 4) interviews, 5) direct-observations, 6) participant-observations, and 6) physical artifact data collection in this study (R. K. Yin, 2011). In this study, we used primary data such as interviews with key informants, secondary data recorded within the company, and in-depth participant observation wherein the first author is a senior employee of the company. This study was conducted between 2019 and 2021, a three-year longitudinal study. Interviews had semi-structured questions and structured questions. The semi-structured questions were devoted to eliciting wide-ranging responses, and structured questions were used to understand certain concepts in terms of importance (Rowley, 2012). We also examined secondary data, such as company internal records, the company data on the process, annual reports, etc. Besides, data on company intranet was also used after taking necessary permission from the company. In order to maintain the confidentiality of the company, the name of the company is not disclosed.

3.2. Analyse the case data

There is no standard cookbook-type recipe for analysing the case data. A good case study research should make use of all data available within the context and purpose of study. Besides, a good case study should consider all rival interpretations and analyses (Rowley, 2012). Besides, it should also try to reduce bias and bring in objectivity to the study. In this study, we followed an explorative and interpretative research strategy. This was carried out in a single case study design using a holistic approach. Thus, we followed a general strategy of working from the ground up. Besides, we searched for patterns, concepts, or new insights, which can come from manipulating the data to understand the boundary between the context and phenomenon. We also used field notes, diagrams, memos used by the main author while carrying out the participant observation in the company. This helped us to understand the phenomenon in an objective manner. As our approach here is to use explorative single case study approach, which will enable us to understand the explanation building with respect to the application of LSS in the diamond plant context. Yin explicates “To explain a phenomenon is to stipulate a presumed set of causal links about it, or how or why something happened. The causal links may be complex and difficult to measure in any precise manner”. Thus, we venture out to examine how LSS is implemented in real life to improve the performance of DMS. In doing so, we also explored the causal links between the factors impacting DMS and how it can be improved. The methodology is depicted in Figure 2

Figure 2. Single case study approach used in this study.

4. The case: background of the company

The diamond corporation is a fully owned subsidiary of the Government and a private group. Each year, it extracts about 1 million carats. The diamond mining and restoration operations and services were established in 1994. The operations are concentrated in Namibia’s southwest coast, with land-based activities concentrated near Oranjemund and additional activities taking place around Lüderitz and along the Orange River. It has approximately 1800 employees. It is a diamond mine situated along the Coast of the Orange River in the Karas region. It has two Plants in the Orange River area. The Recovery plant is where the final concentrate is fed through X-Ray machines for concentration to prevent sending waste to the product stream. It also has plants that are used for exploration purposes. This study was conducted in one of the plants, and the process flow diagram is explicated in Figure 3.

Figure 3. Process flow diagram of the DMS.

Dense Medium Separation is a process that is used to separate DMS concentrate from tailings. Nowadays, dense medium separation of diamondiferous ore is carried out almost exclusively with the DM cyclones. The mixture of heavy medium and ore is introduced into the cylindrical feed chamber of the cyclone tangentially (at an angle). This process results in a vortex or whirlpool-type action. The constant inflow of new feed displaces the rotating mass away from the feed inlet towards the spigot. The heavier or denser particles move to the walls of the feed chamber and cone, which tapers down to the spigot, where they are discharged with the underflow medium whilst the light or less dense particles move to the centre of the vortex. This is because of the restriction caused by the ever-diminishing diameter of the cone section; these, together with the overflow medium are then sucked through an open-ended vortex finder, which runs down the centre of the cyclone. In this manner, the heavier or denser minerals are continually separated from the lighter or less dense minerals. Problems which result due to this process include fluctuations in the correct medium density, cyclone feed pump pressure, viscosity and percentage fines in the feed ore. Density fractionation (the sink-and-float process) is used throughout mineral processing laboratories to ascertain the efficiency of density-based separators. It has also been used as a yield predictor in the coal and iron ore industries, where the density profile of the crushed run-of-mine (ROM) ore is correlated with the plant yield.

4.1. Interviews

The main objective of the qualitative study was to investigate the factors affecting the Dense Medium Separation performance and which factors will have the most significant impact on DMS performance in the company. No sampling technique was used as there were few participants. We asked supervisory and managerial staff to participate, and everyone who accepted were interviewed. Data were collected from all metallurgical managers and engineers. The total population of these staff were 28. Interviews were conducted with 17 employees who agreed to participate in the study. The factors cited by respondents are operational conditions (cyclone pressure and medium density), medium composition (viscosity, clay, FeSi PSD, medium to ore ratio), equipment geometry (spigot and vortex finder) and lastly feed characteristics. The interview questionnaire is attached in Appendix A. In order to find out the most important factors as per the respondents, these factors were ranked based on the frequency in the interviews and are explicated in Table 1.

Table 1. Frequency analysis of factors that impact DMS performance

Factor	Frequency	Percentage (%)
Cyclone Pressure	17	100
Medium Density	17	100
Viscosity	15	88
Cyclone Worn Parts	13	76

The interviewees also cited those operational conditions (cyclone pressures and medium densities), which were the topmost factors affecting DMS followed by viscosity and lastly equipment dimensions. The study then ventured to improve these four characteristics using LSS.

4.2. Improving cyclone pressure

Figure 4 indicates histograms of the cyclone pressures comparing performance before implementation of DMAIC. The detailed DMAIC analysis conducted on the cyclone pressures is explicated in Table 2.

Figure 4. Cyclone feed pressure before DMAIC and lean interventions.

Table 2. DMAIC analysis on cyclone pressure

	Define	Measure	Analyze	Improve	Control
Cyclone Pressure	Cyclone pump pressure fluctuations due to medium instability caused process interruptions leading to potential product loss	Histograms, capability analyses were used to determine variance from set point target to determine process capability. Determine Spiking recoveries	Analyse downtime experienced due to pressure fluctuations	Replace the PC100 densifier with a PC200 to improve the purging process in the DMS circuit by efficiently removing ultra-fines in the DMS. The second suggestion was to install a Variable Speed Drive (VSD) on the cyclone to better control the cyclone pressure to its set point to ensure pressure trends lie within limits and the installed process interlocks on the DMS cyclone pressure feed pump	To install process interlocks on the cyclone feed pump to ensure pressures lie within the limits. Design Standard Operating Procedures (SOPs) to monitor cyclone pressure.

Subsequently, Lean methodology analysis was carried out and is depicted in Table 3.

Table 3. Lean waste due to cyclone pressure variation

Wastes	Consequences of cyclone pressure variation
Defects	Premature failures on the cyclone feed pump impellers, Veins that lead to unplanned plant stoppages
Waiting	Pressure must be stabilized before plant can be fed
Intellect	Operators trained 2 levels up to correctly adjust underflow and overflow valves to better stabilize the pressure to ensure VSD operates optimally

These measures were implemented, and the improved results are depicted in Figure 5.

Figure 5. Cyclone feed pressure after DMAIC and lean interventions.

Before the PC200 was installed the PP (Process Performance index) was found to be 1.03 and Performance Centering Index (PPK) was 0.5 and after the PC200 was installed the PP was found to be 1.53 and PPK was 1.11. The PP and PPK play significant roles indicating how well the data fits between the limits (HCL and LCL). HCL and LCL are the control limits of the process within which the data falls and is critical for process stability and steady state that a process operates within.

4.3. Correct medium density

Figure 6 indicates capability analysis of the correct medium densities before application of DMAIC and lean interventions.

Figure 6. A. Medium density before DMAIC and lean interventions. B. Medium density after DMAIC and lean interventions

The Tables 4 and 5 indicate the DMAIC and lean analysis of the medium density.

Table 4. DMAIC analysis on medium density

	Define	Measure	Analyse	Improve	Control
Medium Density	Medium density fluctuations due to medium instability which caused process interruptions resulting in the medium density dropping which may lead to potential product loss through the cyclone	Histograms, process capability studies to determine variance from set point target to determine process capability Determine spiking recoveries	Analyse downtime experienced due to density fluctuations	Replace the PC100 densifier with a PC200 to better improve the purging process in the circuit by efficiently removing ultra-fines in the circuit To install the process interlocks based on the medium density limits	To install process interlocks on the correct medium density to ensure density lies within the limits. Design SOP to monitor cyclone density

Table 5. Lean waste due to medium density variation

Wastes	Consequences of density variation
Waiting	Density has to be stabilized before plant can be fed.
Intellect	Operators trained 2 levels up to correctly adjust underflow and overflow and how to densify

After the measures were implemented, and the capability analysis of medium density is depicted in Figure 6.

Before the LSS was carried out the mean medium density was 2.49 t/m³ and after the intervention the mean was 2.58 t/m³. It can also be observed that the density trend is much closer to the target of 2.60 indicating that the density trend is much more stable. Before the PC200 was installed the PP was found to be 0.38 and PPK 0.32 and after the PC200 was installed the PP was found to be 1.21 and PPK 0.75. An improvement has been observed after the DMAIC was carried out and applied.

4.4. Improving viscosity related delays

There were a lot of DMS operational delays at the treatment plant prior to the installation of the PC200 densifier. We observed that there were a lot of material-related delays that affected the DMS. Looking at the delays, namely reduced feed rates, viscosity and sandy material, it is noted that all are material related delays caused by the excessive composition of fines. Due to excessive fines in the feed to the DMS, this resulted in a direct impact on viscosity, which also caused the 3^rd delay. It is known that the ore body contains a significant number of fines and clay hence creating operational challenges. After implementation of SOP and PC200 densifier it was found that a significant improvement resulted as depicted in Figure 7.

Figure 7. Down due to viscosity related delays.

Further, it was observed that in the year 2018 and 2019 delays of 174 hours a year were experienced but this number was brought down to 11.5 hours in 2020, an improvement averaging 80% on plant run time. In 2021, the data runs up to April 2021, a quarter of the total hours, and reduction of delays are further observed. Overall, the downtime delays due to viscosity improved, that is, more production taking place, which ultimately reduces cost per ton resulting in generation of more revenue for the organisation. The viscosity-related delays decreased by 136 hours after the PC200 densifier was installed, averaging 80% improvement. A total of N$ 8,421,130 was saved, which is a significant gain for the company.

4.5. Reducing plant down time due to worn cyclone parts

Before DMAIC was implemented in March 2020, the process had no proactive measurements to predict the cone wear. Figure 8 depicts the down time due to worn cyclone parts. A run to failure approach was used as there was no thickness measurement device to predict cone wear. Following the implementation of the DMAIC, it came out that the cone was most vulnerable to being holed, it was agreed that the cone would also be measured using a thickness tester (Q1A—Q5C) and the DMS tonnes processed would start to be measured upon the installation of a new cyclone. Although the maximum limit of the spigot ID, vortex finder ID and length is 10% wear, it was thought best to implement a 68% maximum wear on the cone. Sixty-eight percent wear translates to a final cone thickness of 8 mm from an original 25 mm thickness. It was also known from previous replacements that a cyclone treats about70000tonnes for a cyclone to get holed.

Figure 8. Down time due to worn cyclone parts.

The cyclone was installed on the 7^th of March 2021. Due to coronavirus, the plant did not operate in April, May and June. Thickness testing was conducted every week. On the 23rd of July 2021 the total tonnes treated were found to be 56790 tonnes, which speaks to the thickness test results showing little life to reach the wear limit. The cyclone was eventually replaced on the fourth of August 2021 during service and was proactively replaced causing minimal downtime.

4.6. Validation of the results using spiking of tracers

Spiking of tracers is used to determine the DMS efficiency and recovery efficiencies. Spike tracers have the density and luminescence similar to that of a natural diamond. Spike tracers are introduced in the DMS Feed preparation screen whereby they pass through the cyclone and are transported via a silo to recovery where they pass through X-Ray machines. Spike stimulants of various sizes (4 mm, 8 mm, 16 mm and 20 mm) are used to test the X-ray machine efficiencies while feeding gravel. Spiking tracers have properties like that of a diamond, that is, density and luminescence. The spiking test will provide quality assurance by ensuring diamonds will not be lost to tailings. The metallurgist will introduce the spikes at the feed prep screen, while gravel is transferred into the different size bins. Figure 9 indicates month to date spiking recoveries throughout 2019. The overall column in Figure 9 depicts average value. The target value set by the company is 95% recovery. It can be seen that for 4 mm and 8 mm the target is not achieved. We do observe that there were several spikes missing in the 4 mm, 8 mm and 16 mm size fractions, which indicates potential product loss.

Figure 9. Month to date spiking recoveries 2019.

Looking at Figure 10 we do observe an improvement in spiking tracers’ recoveries after the PC200 densifier was installed in July 2020. We observe that there were several missed spike tracers in the 4 mm size fraction since purple coloured 4 mm spike tracers had visibility issues (low) in the final stage (Sort house) where diamonds and tracers are hand sorted.

Figure 10. 2020 spiking month to date recoveries.

Looking at Figure 11 we observe an improvement in the 8 mm, 16 mm and 20 mm. The 4 mm still had some problems with the visibility, but in 2021 a colour change from purple to green was made.

Figure 11. 2021 month to date spiking recoveries.

Looking at Figure 11 we observe an overall improvement in 2021 compared to year 2020 and 2019. We observe a 100% recovery in the 12 mm, 16 mm and 20 mm. With regards to the 4 mm spike tracers, however, some concerns are still evident in March and May. This is receiving attention but is not a cause for much concern since most of the carats are found in the bigger sized fraction spikes as that is where most of the revenue is generated and it is also imperative for the business growth going forward to ensure profitability and sustainability of mining. This depicts the LSS implementation of DMS has worked.

5. Discussion

LSS is one of the most widely used operational improvement philosophies (Karunakaran, 2016). However, the application of LSS in diamond mining industries has been scarce. This study applies LSS to improve DMS performance in the diamond mining industry in Namibia. The enablers or factors impacting the DMS performance were explicated in this study, and later the most important factors were identified. The key factors unearthed in this study are 1) Cyclone Pressure, 2) Medium Density, 3) Viscosity and 4) Cyclone Worn Parts. The findings were consistent with a study by Malhotra (2009) which found out that cyclone operating conditions, medium composition, equipment dimensions and feed characteristics affect DMS, thus confirming its importance at the Namibian diamond mine. Further, this helped in understanding the key factors to be controlled and their importance to the overall diamond recovery process. It helped in creating awareness about these key factors among key stakeholders, so that the process could be stabilised. Also, these factors were identified as requiring action to solve the problem. These factors were subjected to a DMAIC analysis and Lean waste analysis. The analysis suggested that all four factors could be controlled by a unique solution of replacement of PC100 densifier with a PC200. Furthermore, unique interlocking mechanisms and SOPs were designed. This helped in the implementation of the proposed solution. To make the proposed changes sustainable in the organization, the relevant stakeholders of DMS operations were motivated, explained to and data-based improvement was shown using the results of spiking of tracers. This was important because this would help the organization to implement LSS in the diamond mining company. By understanding these non-value-added activities within the process the flow in the diamond company is improved by elimination of wasteful activities and improving the value-added activities. This is in line with in the TSEF principle of reducing variability (Devaraj et al., 2013). Further, this study revealed that the company should fully utilize LSS within the operations for each process plant to achieve a reduction in downtime delays, an increase in cost saving, enhanced DMS efficiency and reduced waste. By identifying these bottlenecks and remedying the bottlenecks, the materials within the process can move very swiftly and this as per TSEF can improve the flow within the process, leading to productivity (Schmenner, 2004). Also, radio frequency identifiable density (RFID) tracers should be introduced at the company to determine DMS recovery efficiencies and to ensure results are received much quicker and process changes can be rectified at a much quicker rate without the need to take feed off. Further, process owners at the company should be willing to facilitate the implementation of LSS in the different plants. Another suggestion given to the company was to keep the DMS stable by maintaining steady state conditions whereby the process has minimal variations and less waste generated. From a theoretical perspective, this study has contributed to TSEF theory, by applying LSS in a diamond mine. Further, it has suggested how LSS can be applied to reduce variability and improve process flow within a diamond processing plant.

6. Conclusions, limitation and scope for future research

Lean Six Sigma is a widely used philosophy to minimize variations and eliminate waste in manufacturing and service processes. However, its application in mining has been limited. In this study LSS was successfully implemented in a diamond processing plant to improve the performance of DMS which is a critical component in the diamond mine. The study found that LSS implementation helps to achieve cost reduction, enhanced DMS efficiency, reduce waste and downtime delays in diamond plants. This study extended TSEF theory by suggesting how variability within the diamond mining process can be arrested by the application of LSS. Further, the study suggests how various bottlenecks within the process can be reduced by using LSS, thus further extending TSEF in a diamond mining context.

The limitation of this study was that it was a single case study. Besides, due to Covid-19, the plant was operating at reduced capacity sometimes. Future studies should implement LSS in other processes of diamond recovery. Besides, studies should also examine the use of modern technologies such as RFID Tracers to determine DMS recovery efficiencies to ensure results are received online and process changes can be rectified at a much quicker rate without the need to take feed off. To the authors’ knowledge, this was the first study on the application of LSS in a diamond plant to improve the DMS efficiency. Research may also be directed at developing an LSS framework, which should be implemented using multiple case studies to further confirm the usefulness of the framework.

Acknowledgements

This paper is based on Mr Abisai Kanyemba Master of Industrial Engineering thesis guided by Prof Godfrey Dzinomwa and Dr Michael Sony. The authors would like to thank the management of NUST and the Diamond company for supporting this study. We thank the reviewers for the constructive suggestions, which have improved the quality of the paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors received no direct funding for this research.

Notes on contributors

Godfrey Dzinomwa

Godfrey Dzinomwa This study is based on Master of Industrial Engineering thesis of Mr Abisai Kanyemba, which were jointly supervised by the second and third author. The authors conduct research in LSS mining sector.

Appendix A:

Interviews Questions

Interviews for Management

Lean Six Sigma strives to enhance the quality of process yields by recognising and eliminating causes of defects (errors) and reducing variation. More appropriately, Six Sigma is a problem-solving approach that uses data, measurements, and statistics to recognise the essential few determinants that will vividly reduce waste and defects whilst increasing foreseeable outcomes, customer satisfaction, profit, and shareholder value.

Section 1: Employee Information

Name

Designation

Section 2: factors effecting the DMS at Namdeb

Question 1: What factors affect the DMS performance?

Table

Factor	Description	Yes/no
Medium Composition	Particle shape Particle size Clay contaminate Non mag FeSi PSD
Feed Characteristics	Particle shape Particle size ROM composition Near density material
Equipment Geometry	Spigot diameter Vortex finder length
Operational Conditions	Cyclone pressure Medium Density

Other methods to be explained below… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …

Question 2. Which factor has the most significant effect to the DMS performance.… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …… … … … … … … … … … … … . … . … …

Question 3: Do Process Treatment Plants at Namdeb make use of Lean Six Sigma?

Yes No Other

For other methods to be explained briefly below

Question 4: What Quality tools do we make use of at Namdeb ?

Tick yes or no and explain the use at your operations s

Table

1. Control charts
2. Root Cause-and Effect Analysis
3. Histograms
4. Pareto Analysis
5. Scatter Diagrams
6. Failure Mode analysis
7. Process Control Charts
8. PDCA

Appendix D:

Questionnaires

Sample Survey Questionnaire: Dense Medium Separation

Employee Name … … … … … … … … … … … … . … . … …

Signed … … … … … … … … … … … . … . … . … . … . … . … . … . … .

Designation … … … … … … … … … … … … … … … … … .

Name of Plant … … … … … … … … … … … … … … … … …

Date … … … … … … … … … … … … … … … … … … …

Question 1

What factors affect the DMS efficiency?

Question 2

What factor has the most significant impact on the DMS?

Question 3

What are the biggest challenges affecting your DMS Plant?

Question 4

How does the DMS process prevent process variations and what controls are in place?

Question 5

How do you incorporate Lean Six Sigma into your process?