Demand Driven Material Requirements Planning (DDMRP): A Systematic Review and Classification

Purpose: Demand Driven Material Requirements Planning (DDMRP) aims to deal with variability by adjusting inventory levels while maintaining, or even increasing, customer service levels. This approach bridges the push and pull approaches. Even though it first made its appearance in 2011, research in this field remains relatively limited. This paper aims to analyze the temporal evolution of the DDMRP, its context of implementation, and the research themes studied in that field to identify areas that still need to be addressed by future researchers. Design/methodology/approach: The systematic literature review approach used in this paper examines research dealing with the DDMRP published in different languages between 2011 and 2020. To date, published papers focused on performance analysis and comparison of the DDMRP with other methods. This study focuses on the DDMRP itself and its capacity to answer the different operations management functions under a broader vision (context, implementation, setting, etc.). Thus, we analyze the evolution of the method in the literature, considering different languages, and present a taxonomy of published scientific works. Then, we identify gaps that require further research. Findings: DDMRP is not sufficiently studied in the literature and the proposed tools still need to be tested. More research is required to evaluate its adaptability in different and complex environments, address its ability to cope with product variety, complex BOMs, and other methods. A proposal of new approaches to define and adjust DDMRP parameters and evaluate their sensitivity is also required to make the method more scientifically sound. Originality/value: Based on the analysis of papers published in different languages, this paper outlines the current state of the art of DDMRP, its shortcomings, and identifies further research to make DDMRP more robust from both a scientific and industrial perspective.


Introduction
In the past decade, the supply chain has become more competitive, constrained, and evolutive (MacCarthy, Blome, Olhager, Srai & Zhao, 2016). Gollamudi (2013) confirmed that companies must become demand-driven because (I) markets are volatile, (ii) demand fluctuates, (iii) products are specialized, (iv) products have greater variety, (v) lowcost facilities are essential, and (vi) external focus. Thus, companies have tried different and/or new methods, tools, and approaches to reduce their costs and become more flexible and reactive.
Traditional manufacturing planning and control systems were not developed to work in such a volatile context (Kortabarria, Apaolaza, Lizarralde & Amorrortu, 2018). However, developing demand-driven approaches has been the goal of many operations' management researchers in the past two decades. Among them, Grubbström, Bogataj, Bonney, Disney and Tang (2004) reviewed the field of Materials Requirements Planning (MRP) and Production and Inventory Control (PIC) theory and noticed that MRP theory requires information about future demands. In contrast, there is not always such a requirement with PIC theory. Stevenson, Hendry and Kingsman (2005) also presented a review of Production Planning and Control (PPC) with a focus primarily on Make-To-Order (MTO) production. The diversification and shorter production cycle times encourage manufacturers to shift their production from Make-To-Stock (MTS) to MTO or an intermediate system (Kuroda & Takeda, 1998;Stevenson et al., 2005). Tang and Grubbstrom (2002) also investigated a method for planning the Master Production Schedule (MPS) under stochastic demand. They aimed to evaluate the value of replanning actions (assuming the lower-level schedule change cost is known). Salmela and Huiskonen (2019) also developed a new co-innovation toolbox to create a shared understanding of demand-supply chain synchronization between the customer and the supplier. The toolbox includes Demand Visibility Point-Demand Penetration Point, Supply Visibility Point-Supply Penetration Point, and Integrative Synchronisation tools.
With the same objective, Ptak and Smith (2011) proposed applying a new method called Demand Driven Material Requirements Planning (DDMRP). The DDMRP approach gathers features from existing methods and incorporates new, innovative features to manage the material flow. This method aims to address variability and adjust inventory levels while maintaining, or even increasing, service levels to the customer. In turn, this simplifies the job of planning material requirements and improves information flow and visibility (Ptak & Smith, 2011).
The DDMRP was created to resolve certain problems that have been encountered in previous methods. In this context, the nature of MRP, which makes everything dependent, creates nervousness. Thus, the DDMRP method is based on the works of Joseph Orlicky, who proposed the creation of the bill of materials (BOM) in 1965 (Orlicky, 1975) and the concept of the decoupled explosion to establish independent dependence. In the DDMRP case, decoupling points, defined as buffers, ensure that not all changes at a high-level of a BOM are translated into changes in demand of low-level items. Therefore, material flows are somehow protected against the system nervousness that is transferred and amplified in conventional MRP (Ptak & Smith, 2016). Moreover, the DDMRP decoupling point buffers ensure reliable availability of the stock to the consumers, while at the same time allowing for the aggregation of demand orders, creating a more stable, realistic, and efficient supply signal to suppliers of that stock (Ptak & Smith, 2016). DDMRP is conducted in five steps: • Strategic Inventory Positioning: Consists of determining the position of the decoupling points in the supply chain, which act as a variability absorber. • Buffer Profiles and Levels: The buffers are sized to protect flow.
• Dynamic Adjustments: The buffers are dynamically adjusted according to several parameters (Decoupled Lead Time, Lead Time Factor, Variability Factor, Average Daily Usage, Plan Adjustment Factors). • Demand-Driven Planning: The flow is pulled with demand-driven planning, which enables the supply orders to be generated; and • Visible and Collaborative Execution: The execution step manages the open supply orders to protect and promote efficient flow across the supply chain.
As of now, the method has received attention mainly from the professional supply chain management (SCM) and operations management (OM) communities. Some practitioners consider DDMRP to be a revolutionary approach that can result in significant industrial performance improvements, but little public data confirms such a statement.
At the same time, a few papers and a thesis on DDMRP have studied the different elements of the method. However, the method has not yet attracted many researchers. As such, this paper aims to evaluate the status of the DDMRP method in the scientific community by conducting a literature review using a systematic research methodology (Tranfield, Denyer & Smart, 2003). The paper intends to illustrate a clear vision of the DDMRP's progress in the past and to identify gaps that need to be addressed to validate and improve the method using formal scientific methods. Thus, this paper answers the following research questions: • Question 1: How has the DDMRP method evolved since its appearance?
• Question 2: What are the contexts in which the DDMRP method has been implemented or tested?
• Question 3: What are the different research themes that have been studied in the DDMRP literature?
The remainder of the paper is organized as follows. First, the research protocol is presented in Section §2. Next, the analysis framework is discussed in Section §3. Then, we present our results in Section §4, which will be discussed in Section §5, before concluding in Section §6 with the current limits and shortcomings of the DDMRP method as well as proposing future research in that field.

Research Protocol
To answer the research questions, we have collected papers dealing with the DDMRP using the following research protocol.
• Databases: Several databases were initially used, including JSTOR, ScienceDirect, Hal, Taylor & Francis Online, IEEE Xplore, SpringerLink. However, only four databases were retained, as the other ones did not return any other papers on DDMRP than the ones already included in the following databases: • • Query: We attempted to be specific when selecting keywords to collect relevant papers dealing with the DDMRP approach as their main subject of study. To avoid collecting articles mentioning DDMRP but not being the primary research topic, we restricted the search of keywords in the title, abstract, and keywords sections only. Search keywords used were: • "DDMRP" OR "Demand-driven MRP" OR "Demand driven MRP" OR "DDS&OP" OR "Demand-driven sales and operations planning" OR "Demand-driven sales and operations planning." • Publication time: From 2011 to 2020.
Since our study aims to extend the field of analysis, we were interested in collecting and analyzing papers in different publication languages. This explains our choice to research in Google Scholar, as it contains publications in all languages. We used the "search query" to widen the search field to "Full text," as only two options in the advanced search are offered by Google Scholar (either "Full text" or "Title"). We then used the filter based on language.
Figure 1 describes the detailed research protocol used for this systematic literature review. After collecting the articles from the different databases, we removed duplicates and analyzed abstracts and conclusions. The filtering process allowed us to exclude papers on "DVMRP" or "ODMRP", which stand for "Distance Vector Multicast Routing Protocol" and "On-Demand Multicast Routing Protocol", non-scientific publications, and papers dealing with "demand-driven" issues without any relationship to an MRP approach.
At the end of the exclusion process, 57 documents (including Ph.D. and masters' theses) in all languages remained, including 21 papers written in English and 36 in other languages.

Analysis Framework
As there is no precise classification method related to the DDMRP method, we develop our classification schemes to answer the research questions presented in Section §1. The proposed analysis framework includes three main components, and each one is directly linked to one of the research questions. Within each classification component, we define sub-components that allow us to answer each question from different angles. These sub-components represent the elements evaluated within the framework of the DDMRP to broaden our scope of analysis, precisely identify the degree of relevance of the existing works, and identify all the gaps or possible avenues of research.

Component 1: The Evolution of the DDMRP Method
Before starting the analysis of the foundations of the DDMRP method, it is essential to understand how the DDMRP has evolved in the scientific community and to measure the level of interest generated by this new approach. To achieve this objective, our first component compares the evolution of the DDMRP to other methods in terms of time and publication languages. The level of interest is defined here as the number of works published since the method's appearance. Moreover, to evaluate all works conducted in this field, we decided to extend the field of analysis to include any publication language. The first part of the analysis framework is divided into two sub-components: • Sub-component 1-1: The evolution of DDMRP over time. As DDMRP first appeared in the book of Ptak (partner with the Demand Driven Institute) and Smith (co-founder of and partner in the Demand Driven Institute) in 2011, we have chosen to research over the past 9 years (from 2011 to 2020). The first analysis aims at comparing DDMRP evolution against other methods (DDMRP, ConWIP, Kanban, Lean Management, Theory of Constraints, DRP, DDS&OP). The analysis of other methods was conducted by looking at the occurrence of titles.
• Sub-component 1-2: The evolution of DDMRP based on the publication languages. This sub-component analyzes the various works on DDMRP published in different languages to guarantee that all studies are included and to assess the degree of spread of DDMRP throughout the world.

Component 2: The Implementation of the DDMRP Method
The second component studies the contexts in which DDMRP has been tested or implemented. This component aims to identify whether the method has been implemented and adapted to different manufacturing contexts. As such, we propose analyzing, for each work, the characteristics of the implementation context described by researchers. This includes the scope of the implementation method, the methodology used, and the manufacturing operations' specific characteristics studied. To provide detailed answers to those questions, we analyzed the following elements: • the type of case study (pedagogical work, industrial case study).
• the kind of industry (continuous process industry and discrete process industry).
• the workshop type (flow shop, job shop, unique machine).
• the product type (a single product or multi-product).
• the Bill of Material structure (a simple or complex BOM).
• the implementation methodology used.
• the tools and approaches used to study the DDMRP method; and • the presence and use of other operation management methods.

Component 3: The Research Themes of the DDMRP Method
The third component looks at the DDMRP method itself. At this level, we more precisely define the nature of the works carried out within the framework of the DDMRP and the scientific contribution of each work. The purpose of this component is to assess the level at which the work carried out in the literature has sought to discuss, validate, improve, and or replace the original elements proposed by the founders of the method. This component also identifies the limits or gaps that future researchers' studies must fill.

Results and Conclusions of the Analysis Framework
This section presents the results of the literature review based on the proposed classification.

Results of Component 1: The Evolution of the DDMRP Method 4.1.1. Sub-component 1-1 Results: The Evolution of DDMRP over Time
We can observe in Figure 2

Sub-component 1-2 Results: The Evolution of DDMRP Based on the Publication Languages
We can deduct from Table 1 that DDMRP has attracted researchers worldwide, and papers have been mostly published in English, French, Spanish, and Chinese, with a lesser amount of publications in Korean, Italian, and Turkish. However, Spanish and Taiwan researchers tend to publish in their native languages. Cases that have been reported have emerged in various countries, which means there have been a broad range of organizational contexts.
When analyzing all texts in all languages (Table 4 in Appendix A), we note that most papers have addressed parameter and performance analysis of the DDMRP method, as shown in Figure 3. A substantial number of works have also presented a comparison between DDMRP and other methods such as Kanban (Miclo, 2016

Results of Component 2: The Implementation of the DDMRP Method
Concerning the implementation DDMRP method, we deduce from Table 2 the following results: • Case study: Papers based on pedagogical case studies are more present than those on industrial ones. Nevertheless, a few industry cases were studied. For instance, Shofa and Widyarto (2017) and Shofa et al. (2018) presented a study in a manufacturing company in Indonesia to evaluate DDMRP and MRP's performance in terms of average inventory level. Pekarčíková et al. (2019) aim to extend the knowledge base in the area of demand-driven supply logistics in the context of Industry 4.0 and verify the processed theoretical knowledge in a case study.  analyzed the implemented changes and the subsequent qualitative and quantitative results of a company after converting from MRP to DDMRP. Ihme and Stratton (2015) evaluated DDMRP in improving a company's performance and its potential to improve system stability and product availability. Kortabarria and Elizburu (2018) described the student's experience in implementing DDMRP in a company and comparing MRP and DDMRP of the obtained results.
• Field of activity: Industrial case studies come mainly from the automotive (Shofa & Widyarto, 2017;Kortabarria & Elizburu, 2018) and ink production (Ihme & Stratton, 2015) sectors. However, papers written in a language other than English covered other industries. As such, Bahu et al. (2019) and Bahu et al. (2018) analyzed 30 case studies of different organizations and observed that there are no restrictions on the implementation field of DDMRP. This paper suggests that the DDMRP can be implemented in industrial enterprises, but results (failures or successes) are rarely published in scientific journals.
Confidentiality issues might explain part of this lack of dissemination, but cannot be the sole reason, as researchers have overcome these problems with other methods in the past. Thus, more studies deserve to be conducted and published.
• Supply chain segment: The DDMRP method is mainly used to manage the production entity of the SC, except for Erraoui et al. (2019), who attempted to implement the same logic for DDMRP in the distribution segment.
• Level of decision-making: DDMRP is used to manage activities at the operational level. Martin, Baptiste, Lamothe, Miclo & Lauras (2018b) described the DDMRP model's evolution toward the demand-driven adaptive enterprise (DDAE). This model involves a complete set of business rules, from the strategic level to the execution level. Martin, Lauras, Baptiste, Lamothe, Fouqu and Miclo (2019) proposed a control system for the Demand Driven Sales and Operations Planning process. They proposed the first version of a decision-support system and its associated knowledge base by associating several process control methods into a rule-based system.
• Kind of industry: The literature only presented cases where DDMRP was implemented in discrete process industries. To the best of our knowledge, no study was conducted in the continuous process industry.
For example, the Japanese approach of Lean Management was mainly implemented in discrete process industries (Womack, Jones & Ross, 1990;Womack & Jones, 1996) before it gradually penetrated the service industries and continuous process industries (mining, etc.) (Azzamouri, 2018). Similarly, a possible extension and implementation of DDMRP still need to be tested in continuous processes, such as mining, petroleum, cement, gas, steel, pharmaceutical, textile industries. Future research dealing with the study of DDMRP applicability in various industrial contexts can shed some light on the methods' limits and possibilities and guide practitioners. Thus, studies based on scientific research of this concept could encourage manufacturers to apply DDMRP and reinforce the foundations of the method.
• Workshop type: Studying the performance of DDMRP in different types of workshops has not yet been performed by researchers. Indeed, the workshop's nature results in difficulty, which varies from one configuration to another. The performance of the method can vary accordingly. However, a flow shop configuration seems to be the only one that was studied (Miclo et al., 2015;2016a;2016b;Miclo et al., 2019;Dessevre, Martin, Baptiste, Lamothe, Pellerin & Lauras, 2019).
• Product type(s): Papers dealing with one or multi-products to study the DDMRP method's benefits, but researchers have identified no restrictions to more complex products.
• Implementation methodology used: Implementation guidelines or methodologies have not yet been addressed in the literature. Indeed, Orue et al. (2020) presented a systematic literature review to analyze studies that investigate the standardization of the implementation process of the DDMRP model. They have found no evidence of a standardized implementation process for DDMRP that could maximize its potential. They invited other authors to continue researching and defining a standardized implementation process to improve the DDMRP methodology.
• Tools used to study the DDMRP method: In implementing the DDMRP, works dealing with this concept are based on limited tools and approaches to analyze the parameters, the concepts evaluated, and to quantify the results obtained. These tools are distinguished among the simple ones dealing with simulations based on Excel ( • Coupling DDMRP with other methods: Al-Ammar (2018) presented an implementation of a mixed DDMRP-Kanban system on an electrical cable producer in Lebanon. The author affirmed that for an inventory replenishment system, the Kanban model is more efficient than the DDMRP in terms of inventory reduction and customer satisfaction, but only for a limited number of high runners' items. At the same time, DDMRP is more efficient or more suitable than Kanban for the broadest range of stock items. However, a combination of both systems is of interest in order to optimize the inventory management module. A switch to a demand-driven model is much more efficient than the forecast-driven management system. He claimed that, as the DDMRP helped improve the segregation of duties between departments, it provided live monitoring and quicker reaction time to actual demand, helped to avoid bullwhip effects, allowing for more precise inventory management, and ultimately reduced the whole production cycle's lead time.
From production planning and inventory management points of view, two main approaches exist: push and pull strategies. Push strategies are usually suggested for products with small demand uncertainty, as the forecast will provide a good direction on what to produce and keep in inventory. The famous concepts of material requirements planning (MRP) and manufacturing resource planning (MRP-II) are commonly used in this context. Pull strategies are usually recommended for products with high demand uncertainty. Pull strategies were made famous by the Toyota Production System, which proposes the Just In Time and Kanban Card system. However, both push and pull strategies present advantages and disadvantages. This has prompted researchers to propose hybrid strategies that combine both push and pull concepts, optimizing each philosophy's best characteristics (Al-Ammar, 2018). The DDMRP is a new approach that combines traditional MRP advantages and adopts some of the pull strategy concepts (Ptak & Smith, 2011). Until now, there is no study proving that the DDMRP is helpful in all industrial contexts.
As organizations are usually reluctant to abandon their current systems, it is essential to assess how the DDMRP method can be integrated with other traditional methods. When coupled with other methods, the evaluation of the DDMRP performance could also serve to determine its ideal implementation conditions and scope of use. For example, integrating DDMRP with Kanban, ConWIP, or POLKA methods is an interesting question for researchers and practitioners. Unfortunately, current literature has only tried to prove the DDMRP method's efficiency compared with MRP, Kanban, Theory of Constraints, etc.

Conclusion-Component 2:
The DDMRP method cannot be considered a generalizable method to all contexts because various elements are not considered in current studies. Also, the performance of the DDMRP in a complex context cannot be predicted. Indeed, scientific research in the DDMRP field is mainly based on pedagogical case studies. Thus, more investigation in real industrial contexts is required to assess the method and evaluate its adaptability capability in different and complex environments. Researchers also need to address the method's ability to cope with varying configurations of workshops, product variety, complex BOMs, and coupling with other methods. For example, case studies describing success or failure factors could be shared, as well as specific details of implementation methodologies used. Formal scientific experimentation is also required to quantify the performance of the DDMRP and establish solid scientific foundations.

Results of Component 3: The Research Lines of the DDMRP Method
In the following subsections, we summarize the articles analyzed according to each component of the DDMRP method. Results are summarized in Table 3.

Sub-Component 3-1 Results: Buffer Positioning
Despite its great importance within the DDMRP method, research on buffer positioning is surprisingly limited. Rim et al. (2014) presented a model that determines the optimal position and quantity of a Work In Process (WIP) inventory for a given bill of material using the Actively Synchronized Replenishment (ASR) lead-time with a resolution approach based on a genetic algorithm. In continuity with this work, Jiang and Rim (2016) extended the previous study to the general BOM(G-BOM), in which parts can have more than one immediate parent. Jiang and Rim (2017) addressed the same issue by introducing the stochastic processing times character. Other papers (Shofa & Widyarto, 2017;Shofa et al., 2018;Velasco-Acosta et al., 2019;Pekarčíková et al., 2019) discuss the buffer positioning problem without proposing new approaches.

Sub-Component 3-2 Results: Buffer Profiles and Levels
Ihme and Stratton (2015) evaluated the capability of DDMRP to improve a company's performance. They discussed its potential to improve system stability and product availability by using the original techniques to determine buffer profiles and levels. Shofa and Widyarto (2017) and Shofa et al. (2018) evaluated and compared DDMRP and MRP's performance in terms of average inventory levels in the system. Kortabarria and Elizburu (2018)  However, and to the best of our knowledge, only Lee and Rim (2019) tried to improve the original buffer profile and level method. They proposed an alternative mathematical safety stock formula for DDMRP replenishment. The authors indicated that the proposed formula outperforms the DDMRP guidelines and existing safety stock formulas in average inventory and stockout rates.
Thus, different articles start with a description of the buffer sizing principles proposed by Ptak and Smith (2011;2016) and then move to its application in a pedagogical or industrial case study. Then, the authors tried to quantify the obtained benefits in terms of the primary key performance indicators: WIP, WC (Working Capital), and OTD (On-Time Delivery). Thus, their objective was not to analyze or improve the dimensioning buffer method but to prove that this concept can improve a firm's performance. However, the reported improvements are not compared to the results obtained with other approaches.

Sub-Component 3-3 Results: Settings & Adjustments
The DDMRP method is based on parameter settings that vary for each company. Parameters are linked to a system's specific characteristics and include Average Daily Usage (ADU), Decoupled Lead Time (DLT), Lead Time Factor (LTF), Variability Factor (VF), etc. (Ptak & Smith, 2016). These parameters are not defined precisely by their founders. Surprisingly, we did not find many works proposing new parameter setting procedures. Miclo et al. (2015;2016a;2016b) performed comparison and parameter analysis based on multiple scenarios. For instance, Miclo et al. (2015) dealt with some changes in the lead-time factor. Miclo et al. (2016a) also treated several external variability sources such as spike demand and seasonality. They (2016b) also combined two types of variability: internal (instability of operating times and setups) and external (spike demand and seasonality of demand). Finally, they (2019) evaluated their effectiveness relative to two other widely accepted approaches -MRP II and Kanban/Lean production -through a series of structured computer simulation experiments. The experimental design consists of two factors: type of planning system (at three levels) and demand variability (at two levels). When applying it to a modified flow shop with a competence bottleneck, results showed that the buffer sizes' dynamic adjustment reduces stocks while ensuring a satisfactory service level. They found that their approach increases the workload to an extent where a specific limit must be determined to avoid having lead times drastically increase.
Some researchers also studied the impact of order sizes in replenishing buffers. Most papers deal with the Net Flow Equation (NFE) defined in the basic DDMRP method. The NFE seeks to replenish the buffer levels to reach the top of green (TOG) based on infinite capacity. In a real context, a workshop has a limited capacity, which is defined by its bottleneck. Taking into consideration complex BOM and shared components between different final products is complex and is not studied in the DDMRP literature.
Also, no papers deal with the parameters' dynamic adjustment according to the system's state or particular conditions to be defined. Similarly, no researchers have studied the parameter setting step under loads and capacity variations.

Sub-Component 3-4 Results: Demand-Driven Planning
Researchers rarely study the planning step of the DDMRP method.  and Pekarčíková et al. (2019) have dealt with this step but under a validation objective of the method. Using an illustrative case, they have discussed the importance and advantages of this sub-component compared to a previous system by avoiding the purchase of large and unnecessary batches of the same product.
On their part, Martin et al. (2018b) and Martin et al. (2019) have improved the original planning method. They first proposed Demand Driven Adaptive Enterprise processes based on a series of experts' interviews and the study of existing literature. Martin et al. (2019) then proposed control and decision-making processes for Demand-Driven Sales and Operations Planning. By associating several process control methods into a rule-based system, the authors developed the first version of a knowledge base decision support system.

Sub-Component 3-5: Visible and Collaborative Execution
No papers deal with the visible and collaborative execution step of the DDMRP method. However, order tracking was discussed briefly by . Pekarčíková et al. (2019) also discussed the problem of order prioritization and launching alerts.

Discussion
To evaluate the scientific maturity of the DDMPR method, we conducted a cross-analysis consisting of distinguishing the nature of the scientific contribution of each paper that has been treated before. Our goal is to determine whether they contribute to validating the original method, improving the proposed techniques, or creating completely new tools or methods. Note that we ignored papers that present literature reviews or a simple discussion about the DDMRP method, as these papers do not provide new knowledge on the method itself. Figure 4 first illustrates the current state of the research on DDMRP. We can note that the DDMRP method has as of yet only been partially studied. Indeed, among the 15 possible types of contributions, only 9 (60%) were addressed. Additionally, the figure demonstrates that the research in the field is still not mature, as most research still focuses on validating the original method. The vast majority of papers tend to validate the impact of buffer profiles, levels, and their dynamic adjustments. The latter two steps of the method have not been tackled much, with only six published papers. We also note that new methods, either an improvement in the original methods or new ones, remain rare, with only seven contributions. This analysis confirms that future research is required to pursue the DDMRP method's development from a scientific standpoint.
In our second analysis, we positioned each paper according to its primary scientific approach. Each paper is represented using a colored rectangle with the article numbers (N°) used in previous tables. Both qualitative and quantitative approaches are considered here and are represented by color codes in Figure 5. Our analysis leads to the following observations: • Validation has only been conducted through case studies and simulation. Surveys are still ignored, and quantitative data reporting actual performance measures have not yet been published. This may explain, in part, the lack of interest in the method from the operation management scientific community. • There are intense uses of simulations to improve the various DDMRP components. Metaheuristics, all genetic algorithms, were also used to solve the buffer positioning problem. There is a glaring absence of exact methods such as optimization. This may be explained by the complexity of the manufacturing systems studied and the high number of variables that may influence decisions and system performance. • The setting of buffer profiles and levels is the most studied theme. However, most research is aimed at validating the original method, and only one paper proposed an alternative way of calculating the safety stock formula for DDMRP replenishment. This is also the case with the dynamic adjustments step, with only one paper proposing a dynamical adjustment of the decoupled lead time of the DDMRP based on discrete event simulation. • Researchers have hardly explored Demand-driven Planning. Surprisingly, this is the only theme that has led to new development. • Visible and collaborative execution, the last step of the method, has been largely ignored, and only a partial validation of the proposed was conducted.
These observations suggest that the DDMRP method has not yet reached a high degree of scientific development. As with any new approach, the main priority of the researchers was placed on validating the proposed tools and techniques. Still, validation has not been carried out extensively, as several method elements have not yet been validated, or they rest solely on qualitative analysis or a simulation of simple manufacturing systems. Specific methods of improvement and new developments seem pretty difficult because of the complexity of the problem. Indeed, testing one individual technique or method at a time is difficult, as the method has been designed to operate as a whole.

Conclusions and Perspectives
This paper has presented a systematic literature review of Demand Driven Material Requirements Planning (DDMRP). Our results have shown different shortcomings in the DDMRP method that need to be addressed by researchers.
First, the DDMRP method is not sufficiently studied in the literature, as the number of papers published in this field is low compared to other methods. Articles dealing with DDMRP have been published in a variety of languages. However, many authors published preliminary results in their native language, and only a few fundamental DDMRP papers were published in international journals or conference proceedings (in English). However, these international publications may provide value to researchers and practitioners, as they can serve to test the method in different contexts.
Second, most of the articles analyzed are oriented towards parameter and performance analysis or a comparison between the performance of DDMRP and other methods. Thus, research aiming at improving the method should be encouraged.
Further studies are also required to analyze the impact of DDMRP in real and complex industrial conditions. The uses of the DDMRP method for supporting continuous production processes or organizations using non-flow shop configuration remains untested. Future research must also consider the management of multi-products, which is much more complex than single product settings. A complex BOM structure also deserves to be studied to evaluate the actual value of the DDMRP method when priority problems among components belonging to several products emerge.
Further research is also required to define the limits of DDMRP and how it could be coupled with other existing methods. Indeed, each industry uses specific methods and tools, which may be difficult to fully replace. Hybrid systems, such as DDMRP coupled with ConWIP, may also resolve production order generation and priority management problems occurring in DDMRP systems.
From a practical and scientific perspective, the DDMRP parameters settings (DLT, LTF, VF, ADU) raise some reliability or credibility issues. In-depth analyses of current parameter settings, their dynamic adjustments, and an analysis of their degree of sensitivity and interdependence should be a priority for researchers. Further research is also required to analyze the performance and dynamics of the DDMRP system when dealing with disturbances such as variable demand arrival and varying capacity.
Finally, more research is required to guide practitioners in implementing the method. Indeed, researchers have not yet studied or proposed any implementation methodology. Critical success factors and risk analysis should also be treated to better assist organizations that seek to implement the DDMRP method. As it stands currently, practitioners have to rely on consultants only.
In conclusion, this systematic literature review demonstrates that DDMRP is not yet a mature method from a scientific perspective. More research is required to address its shortcomings or to demonstrate its full capability.

Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.