Can ERP learn from DCS? ERP systems promised too much and delivered too little. They could benefit from the approach used to develop distributed control systems.

Many companies have been swept away by "ERP systems" without really understanding the purpose or consequences of such a move. To begin with, enterprise resource planning (ERP) is a business philosophy, not a software package. ERP is a natural evolution from manufacturing resource planning (MRP II), which evolved from material requirements planning (MRP I).

To efficiently address these business philosophies, companies must install appropriate business systems. Unfortunately, many firms have been sold on the need to roll out a single, all encompassing software package on a single server as the "solution." Part of this problem stems from our attempts to measure and manage everything from one place. CEOs have been sold on the idea that if they buy "Brand X" ERP they will have all corporate data in one system, running on a standard technology set. CFOs will have all the numbers. The systems will be perfect once installed! Often, however, this process has gone astray. Why? Because companies are imposing an architecturally flawed system.

The same "integration dream" was prevalent in the early days of process control. Mills attempted to implement all encompassing control systems. The whole facility was to be controlled by one central computer, from one central control room, by one operator. Unfortunately, as processes became more complex and plants became physically larger, even the biggest control computers were unable to manage effectively. Networks ran out of bandwidth. Humans ran out of coping skills. Systems became unmaintainable.

DCS suppliers and users learned from these early miscues. Modern distributed control systems (DCS) are modular and hence more effective. They are designed to achieve specified levels of repeatability, reliability and repair ability. They consist of a number of linked components, such as smart transmitters, distributed control modules, plant historians, analysis tools, and model-based decision support systems. They work together as an integrated system (see box above).

The approach to "plant control" has evolved. Functions such as local control, inter-area control, and data analysis are clearly distinguished and architected appropriately. No attempt is made to do all things in one system or in a group of modules. There is focus. The result is an "integrated system" with components that do a number of tasks well. My suggestion is that ERP solutions should be architected, selected and implemented in a similar manner.

System features that address repeatabilty

* Measurement devices are self-tuning) assuring performance

* Data is maintained where it is used and stored only as long as needed, reducing network traffic

* Software tracks valid data ranges, and the system completes data reconciliation through sophisticated numerical methods maintaining data integrity

* Bad or suspect information is flagged, to control its use

* Numeric and textual information is linked, to add clarity to data

* Simulators are used to train operators and staff to achieve consistent responses

System features that address realibility

* Redundant components are installed to maintain critical data values

* Systems are designed for 24 hour, seven day a week performance to ensure continuous operations

* Computing "horsepower" is applied where required to manage performance

* Alarm managers and process simulation are used to focus control actions

* Data for control operations ale maintained separately to ensure performance

* Data required for business records and analysis are extracted into a separate historian system where they can be manipulated, charted, alarmed, and reported without interrupting plant processes

* Components from different vendors are assembled and tested against high fidelity process models to ensure achievement of design requirements

System features that address repairability

* Critical components incorporate hardware redundancy and hot swappable components to help technicians complete maintenance and upgrades on a modular basis

* Levels of decision support are standardized to aid troubleshooting

* Integration is managed to control complexity

* Control functions share data only as required to reduce complexity

* Inter-systems communication hooks are defined.

* Integration software and, it necessary, specific integration hardware, are purchased as catalog part numbers, reducing potential variability

* Links to business systems extract only key data using standard tool sets, simplifying upgrades

Bill Thomas is director of business systems, forest industry consulting group, for AMEC E&C Services Ltd. (formerly HA. Simons), Vancouver, British Columbia, Canada. He was with Simons from 1981 to 1993, where he became director of systems engineering and was responsible for integrated millwide and corporate systems. In 1993 he joined BC Hydro as manager, systems planning. In 1995 Thomas joined MB Paper as manager of manufacturing support, a job that included the rollout of a new SAP ERP system. In 1998 he served Port Townsend Paper as director, business systems. In late 2001, he rejoined AMEC. Contact him at: [e-mail]