June 1999
In 1997 more than 20,000 firms worldwide paid $10 billion to enterprise resource planning (ERP) vendors [1]. With a growth rate of 37 percent per year, total ERP company revenue is expected to top $52 billion by 2002 [2]. ERP is expected to continue to be one of the largest, fastest-growing and most influential players in the applications software industry well into the new millennium. Yet little is found about ERP in the operations research literature. The purpose of this article is to begin to fill this gap by making the OR community more aware of how ERP may affect OR practice and research in the future.We will start by reviewing ERP, its characteristics and its stated benefits, drawing parallels with another industry-driven system, material requirements planning (MRP). Next, we will explore potential areas of interest within the broad field of OR before concluding with some general comments and observations regarding the future of integrating OR and ERP.
Enterprise Resource Planning
Enterprise resource planning is a software architecture that facilitates the flow of information between manufacturing, logistics, finance and human resources functions within a company [3] and, as such, is an enterprise-wide information system [4] (See Figure 1). Using a centralized database operating on a common computing platform, ERP system components interact with an integrated set of commonly designed applications, consolidating all business operations into a uniform system environment [5].
In an ERP system, data is entered into the computer system once and only once [4] so that all applications use data that are consistent, complete and common. For example, when a sales representative enters an order into a company's ERP system, the data are immediately available to the plant floor. Manufacturing can then begin fulfilling the order if the item is not in stock, as accounting is formulating the customer invoice and shipping is notifying the shipper of a future delivery. Ideally, the customer can interrogate the system to monitor progress of the order. All is done using the same data and common applications.The integrated ERP applications provide consistency and visibility for all activities across the organization. This consistency is also reflected in a company's data; data fields are defined identically across the entire enterprise [6]. However, to attain the full benefits of ERP, an organization must re-engineer many of its business processes to align with the processes embedded in the system, a very complex set of activities [7]. Before discussing this change, we need to review some of the characteristics of ERP.
ERP systems employ client/server technology [6], allowing a user's (client's) system to run an application (accounting, inventory management, etc.) that accesses information from a common, core database management system (server). This system reflects the concept of decentralized computing simultaneously with a centralized database.
An ERP system includes a number of ERP applications, or modules which typically are the functional software packages for each individual business unit (finance, human resources, order processing, logistics). The current trend in ERP systems is for vendors to offer specialized applications for unique processes and procedures within a given industry. These modules service vertical markets such as government, health care, financial service or retail environment [8]. Also, ERP developers are concentrated into specialized areas, such as supply chain management, demand forecasting, and sales and marketing automation.
Parallels with MRP
Material requirements planning is a set of procedures for converting forecasted demand for a manufactured product into a requirements schedule for the components, subassemblies and raw materials comprising that product. The development of MRP and its various extensions (manufacturing resources planning, distribution requirements planning) led to new dimensions and expansions of OR models, although with a significant delay from MRP's rapid industrial growth to its recognition in the OR community.
In the late 1960s and early '70s, MRP was used to assist in managing inventory and scheduling replenishment orders. It aided in reducing inventory levels, increasing the effective resources and improving overall customer service. However, it was not until the late 1970s that MRP became of interest to the OR community and scholastic papers began appearing in OR journals including Management Sciences, Operations Research and Interfaces (e.g., R.C. Carlson et al. [9]). The research initially reassessed the relevance, validity and applicability of conventional inventory management techniques, such as the EOQ model [10]. Issues such as optimal lot sizing (with time varying demand), alternative lot sizing schemes and heuristics were then addressed as areas where OR could make a contribution. Today, MRP is considered an integral part of OR and is a required topic of instruction in operations management. We suggest that ERP will have similar implications in both fields.
Benefits and Limitations of ERP
A primary issue that concerns all aspects of operations research is data acquiring, reconciling and manipulating data. Traditionally, OR models are created with an implicit assumption that the necessary data are readily accessible. However, such is rarely the case. Normally much of the essential data exists in diverse databases that feed disparate applications which employ non-uniform formats, definitions and structures for similar data items. Before use, these data must be reconciled across multiple interfaces, platforms and databases [6].
Data that cannot be isolated is either ignored in the model, gathered independently of the application, or assumed to follow some standard (well-known) mathematical function. Data collection is tedious, time consuming and costly, and it tends to require the majority of project time when creating a model. Unfortunately, when such models are finally solved, the results, though potentially "close enough" to optimal, often lack practical credibility due to all the data problems.
ERP systems operate utilizing a single set of data across the enterprise [6]. As data is entered into the system only once, all data fields across the organization must be commonly defined. Therefore, all data in the system must be compatible, and thus, more easily accessible as there is only a single database. ERP systems eliminate the problems previously associated with data collection: availability, accessibility and compatibility. This can have a significant impact on the development and use of OR models.
As a result of ERP's ability to provide uniform access to data, model development time is substantially reduced. Since data definitions are uniform throughout the enterprise, there is strong consistency within the data, thus leading to more accurate models and greater credibility of the results. As a result, reusability of the model is increased, permitting shorter set-up times to develop model changes. This also provides the leverage to construct more complex models to include factors and variables that would formerly have been disregarded as being too difficult to collect. By increasing the boundaries of traditional modeling, uniting OR with ERP systems will help lead to deeper understanding of business operations.
Driven by business process re-engineering, the implementation of ERP systems reduces data and process redundancy within an organization. The CIO of Steelcase Inc. remarked that "...we can achieve an $80 million reduction in operating expenses just by getting rid of redundant processes and cleaning up our data"[11].
With elimination of duplicate data, another dimension of data analysis in which OR will play an important role in ERP is data mining (for more on data mining, see page 26). Data mining is the application of advanced OR algorithms, statistical analysis techniques and artificial intelligence techniques to large quantities of data in order to discover hidden trends, patterns and relationships [12]. For example, retail stores can obtain profiles of customers and their buying patterns, and supermarkets can analyze their sales and the effect of advertising on sales [13] to find hidden trends.
To date, only a small part of the full range of possible data mining applications has been explored. Common problems associated with data mining projects include those that face most OR applications: insufficient and unreliable data. In situations where data was available, projects are often not viable given the costs of accessing and preparing the data. However, with the more and better data from ERP, many new data mining applications should be discovered.
One of the key attributes of data mining is that there are typically many contributing factors in any given problem. For example, inventory forecasting can use customer purchase history data, store geographic and demographic data, sales promotion data, and even weather data to help forecast expected demand for product. This kind of data-rich problem is known to have very complex yet subtle relationships among many variables. With its enterprise-wide database, ERP provides a platform for easier access to the data necessary for this type of analysis. ERP will also enable the scope of data mining to expand across the entire enterprise and beyond, encouraging the application of data mining OR techniques over a much broader area. Data mining will become critical to the enterprise that wishes to exploit operational and other available data to improve the quality of decision-making and gain critical competitive advantage.
ERP system vendors recognize that time is a critical constraint for a business, hence they have created methods in the ERP software for reducing cycle times. Bruce Johnson, CIO of Colgate Palmolive Co., said, "The day we turned the switch on we dropped two days out of our order-to-delivery cycle" [6].
With integration between functional areas, the framework for cross-functional communications among unlike business units is more prevalent. For instance, human resources and operations are more likely to communicate. Thus, functional tasks such as project scheduling and staff scheduling could be linked. Using well-known operations research tools such as linear programming, the best mix of employees' skills for any particular project could be determined. Hence, a sort of Just-in-Time project scheduling system could be established, scheduling employees' work hours so that their skills are available only when needed in order to reduce cycle times. This scheduling system could be linked both with purchasing and hiring departments. New personnel could be hired such that their skills reflected those of any "soon to retire" employees as well as new skills that may be required to operate any new equipment set to be purchased. This would expand the scope of traditional OR.
Using OR with ERP allows business decisions to be analyzed enterprise-wide, resulting in time savings, improved control and elimination of superfluous tasks. For example, a year after implementing ERP, Par Industries in Moline, Ill., reduced lead time to customers from six weeks to two weeks, on-time delivery performance increased to more than 95 percent, work-in-progress inventory dropped almost 60 percent, and the life of a shop floor order went from weeks to hours [14].
Often, top management evaluates functional unit performance based on cost savings of individual units during any fiscal period. Thus, each unit independently tries to minimize its costs. This creates a segregated environment where knowledge becomes proprietary to a particular business unit, resulting in a "protecting one's turf" type of sentiment between functional areas [15]. However, as is well known to the OR community, minimizing costs of each unit does not necessarily result in overall cost minimization (or profit maximization) of the entire enterprise. ERP systems will allow OR modelers to conduct total cost minimization analysis over the entire enterprise, thus realizing potential functional cost trade-offs. This could also create new perspective for defining more accurate local and global performance metrics within an organization.
Interestingly, many continuous process industries, such as the oil and gas industry, the telecommunications industry and the airline industry, have real-time optimization performed on a continual basis for obvious reasons to minimize cost, delay and irritations to the consumer. For example, when all Boeing 737s were pulled out of service for maintenance checks, airlines were forced to quickly reallocate their remaining airplanes for minimal disruption and delays in service. ERP systems provide the framework for similar real-time optimization within other manufacturing and service industries, thus opening up a major new opportunity for real-time OR modeling.
Recognizing companies' needs to reduce their time to market for goods and services, ERP systems are designed to respond quickly to new business demands and can be easily changed or expanded without disrupting the course of business. "Companies are always finding new ways to go to market," says Larry Ferrere at J.D. Edwards. "Your business may not always involve the same products. Internally you will have new business requirements, so you have to be positioned for change" [6].
To remain competitive, manufacturers must be able to meet a customer's demand for delivery of product despite any unique specifications, schedule or global location of the customer. When a customer places an order, the firm is often required to provide the customer with a committed delivery time, frequently with repercussions for not meeting this promise. Traditionally, this time frame was provided to the consumer by a customer sales representative, normally with little attention to current inventory levels or production schedule, and thus without knowledge of the true potential order delivery date. ERP allows the sales representative to easily, quickly and simultaneously consider inventory-on-hand, raw material availability, work-in-progress and production capacity across the enterprise. It permits the firm to provide the customer with the highest degree of confidence in their delivery time commitments. This in turn allows for real-time adjustment across the enterprise to accommodate the new demand. Customers are thus integrated directly into the planning process and are provided with the best possible promise date, while supply chain output is maximized.
However, available-to-promise (ATP) dates may also cause the system to be subject to system nervousness too frequent updates that cause overreaction. This occurs when the system is modified too often, and the overall efficiency of the system is diminished rather than increased. System nervousness is typically avoided by establishing arbitrary rules to restrict the number of system updates or by disallowing changes after a certain time period. Unfortunately there are very few models to aid managers in determining when to use which solution or what the exact levels or timeframes should be. This may provide operations researchers with a new class of OR models: ATP management.
Implementation of ERP is a time-consuming, expensive and arduous task. Forty-four percent of Fortune 1000 IT executives reported that they had spent at least four times as much on ERP implementation help than they did on the software license itself [16]. With such an extensive concentration of resources on one system, the OR community may find itself resource constrained and unable to take advantage of the potential that ERP provides.
There is one interesting ERP problem on which OR can have direct impact: deciding when and how often the colossal ERP database should be updated. With many functional units having read and write access to the same data, it will be necessary to establish rules and constraints to decide which business area has the ultimate authority for a particular data set in the circumstance when data entries conflict. This could be a fruitful area of future OR research.
ERP systems force firms to re-engineer current practices to fit within the processes described by the ERP modules. Selecting the wrong software could result in an unwilling commitment to an architecture and applications that do not fit with the organization's strategic goals. Software has taken over the defining role that hardware formerly held. "We used to be an IBM shop. Now we are an SAP shop," said David Edelstein, vice president of information management at Bristol-Myers [17].
The implication of this for the OR community is that existing models will need to be adapted to the ERP system and not the other way around. While legacy OR models will require the ERP data, the typical ERP system will only use the OR models if they have been modified to fit the requirements of the ERP system. As database administrators, systems analysts and systems programmers have discovered, sometimes it is better to simply scrap existing systems than to convert them. The retrofit or re-do will be a major challenge to OR developers and decision-makers.
Letting one vendor provide most or all of a firm's enterprise systems is an attractive but risky proposition. "If you depend on a single vendor, you get a common architecture, lower support costs and cheaper seats," argues one analyst with Gartner Group Inc. "On the other hand, upgrades become a bear because you have to do everything at once" [18].
Such issues become even more important when an OR modeler attempts to investigate supply chain management. Recent developments in ERP systems suggest that a company may link its ERP system with its suppliers' suppliers and its customers' customers. Having suppliers connected to the ERP system suggests that the suppliers will be able to monitor a company's inventory levels for the parts they supply. This could instigate a shift in the responsibility of managing inventories, placing it on the suppliers rather than the company, thus shifting inventory costs. The inventory shift could result in cost savings for the company resulting from the economies of scale in terms of the supplier interacting with several companies for a given part.
The problem with this is that to be successful the firm's ERP system needs to link with their customer's ERP system. As was discovered with electronic data interchange (EDI) and computer-aided design (CAD), customers and suppliers rarely agree as to the best solution. It is not uncommon for a supplier to have multiple versions and suppliers of EDI and CAD systems, each tied to different customers. While ERP may provide OR practitioners with the environment to model and optimize the entire supply chain from key supplier and material availability through production and inventory management to final distribution to the customers it may also generate extensive incompatibilities among members of the supply chain.
Conclusions and Implications
The vendors of ERP systems are realizing the necessity for OR techniques within their systems. Market dominator SAP utilizes a Logistics Optimization Interface (LOI) in its flagship software R/3, which makes it possible to optimize manufacturing processes [19]. Starting mid-1998, SAP introduced its APO (Advance Planning and Optimization) functions. These include functionality for supply chain modeling and monitoring, forecasting and demand planning, available to promise (ATP), and advanced planning and scheduling. In April 1998 the Baan Company, a leading supplier of enterprise business software, introduced a new suite of best-in-class software for demand planning, supply chain optimization, factory planning, scheduling and execution management [20]. Baan also recently acquired CAPS Logistics Inc., a market leader in optimization software for logistics planning and scheduling [21]. PeopleSoft, a California-based ERP vendor, offers its customers a supply chain and optimization solution for their ERP systems. This application package enables planners to balance available capacity while simultaneously taking into account material arrivals and inventory constraints within the company as well as across the supply chain. Other ERP vendors are following suit with similar packages.
There seem to be three natural areas in an organization where OR and ERP overlap: the internal organization, the external organization and organizational modeling. These areas may simply highlight the impact of ERP on traditional modeling. However, they may also lead to the discovery of extensions of existing models or possible new classes of models as a result of ERP, as well as reveal the limitations of some currently accepted techniques. It is important, therefore, that the OR profession take notice of this important change and rapidly show how operations research can both benefit and be a benefit for ERP.
| Roadblocks to Implementation |
Despite the benefits and expanding growth, not all ERP implementation projects
are completely successful. The major causes of less than totally successful implementations are:
Adequately trained users are also critical for the success of an ERP project. It may only take days to change hardware and software, but it takes weeks or months to scale learning curves. The "best" ERP system can be of no benefit if no one knows how to use it.
|
| How to Survive an ERP Implementation |
Keeping ERP projects on track and on budget is difficult. However, there are some guidelines for a successful ERP implementation, regardless of vendor or project:
Complete integration among applications is very difficult to realize. Despite ERP vendor efforts, many companies will not be able to achieve this goal. Thus, they will operate with partial ERP systems, i.e., systems for specific functions such as human resources and payroll, but will not be able to eliminate legacy systems completely. Reference
— Anne G. Robinson and David M. Dilts |
References
- Martin, Michael, 1998, "An ERP Strategy," Fortune, Vol. 137, Feb. 2, pp. 149-151.
- AMR Research, 1998, "Enterprise Resource Planning Software Report, 1997-2002."
- Hicks, Donald A., 1997, "The Manager's Guide to Supply Chain and Logistics Problem-Solving Tools and Techniques," IIE Solutions, Vol. 29, No. 10, pp. 24-29.
- Lieber, Ronald B., 1995, "Here comes SAP," Fortune, Vol. 132, Vol. 7, pp. 122-124.
- PeopleSoft, 1997, "PeopleSoft: Meet PeopleSoft," http://www.peoplesoft.com/ (Dec. 10, 1997).
- Stevens, Tim, 1997, "ERP Explodes," Industry Week, Vol. 245, No. 13, pp. 37-40.
- META Group, 1997, "The SAP Invasion - North American Utilities' Buy-In: Part 1," http://www.metagroup.com/newwhos.nsf/InterNotes/Link+Pages/uits+-+delta2, May 9, 1998.
- McKie, Stewart, 1997, "Packed Apps for the Masses," DBMS, Vol. 10, No. 11, pp. 64-68.
- Carlson, R.C. et al., 1979, "Less Nervous MRP Systems: a dynamic economic lot-sizing approach," Management Science, Vol. 25, pp. 754-761
- Orlicky, Joseph, 1975, "Material Requirements Planning," pp. 3-29.
- Stein, Tom, 1997, "Key Word: Integration," Information Week, Sept. 22, No. 649.
- Datasage "Data Mining?" http://www.cirrusrec.com/Data_mining.html, May 25, 1998.
- Silicon Graphics, "Data Mining with Silicon Graphics Technology," http://www.sgi.com/Technology/data-mining.html, May 25, 1998.
- Appleton, Elaine L., 1997, "How to survive ERP," Datamation, Vol. 43, No. 3, pp. 50-53.
- Bernbom, Gerald, 1997, "Institution-Wide Information Strategies" CAUSE/EFFECT, Vol. 20, No. 1, pp. 8-11.
- 16. Michel, Roberto, 1997, "Reinvention Reigns: ERP vendors redefine value, planning and elevate customer service," Manufacturing Systems, Vol. 15, No. 7, pp. 28-92.
- Weston, Randy, 1997, "Huge enterprise apps pose management risk," ComputerWorld, Vol. 31, No. 45, pp. 10.
- Stein, Tom, 1997, "Enterprise Apps - Not Just ERP Anymore: Some customers want sales, supply-chain, and other links built into their enterprise resource planning products. Vendors are stepping up," Information Week, Dec. 1, No. 659.
- Breuer, Martin, 1996, "Complete Integration," http://www.sap.com/, April 19, 1998.
- Baan, 1998, "The Baan Company Expands Its Advanced Planning and Scheduling Suite and Announces a Focused Business Unit to Provide Affordable Integrated Supply Chain Optimization," http://bvw.baan.com/external/press_analyst/, May 12,1998.
- CAPS Logistics, 1998, "Strategic Acquisition of Leading Logistics and Transportation Planning and Scheduling Software Vendor," http://www.caps.com/newsevents/pr/baan.htm, Jan. 3, 1999).
Anne G. Robinson is a Ph.D. student in the Department of Industrial Engineering and Engineering Management at Stanford University. Her research interests focus on improving the supply chain through information flow, in particular ATP systems.
David M. Dilts is a professor in the Department of Management Sciences at the University of Waterloo, Ontario, Canada. An element of his research deals with the integration of heterogenous operations management information systems.
OR/MS Today copyright © 1999 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com
Web Site © Copyright 1999 by Lionheart Publishing, Inc. All rights reserved.