Integrating process design and control improves plant operability

Considering optimum control strategies and process design together, rather than adding control on top of a finished design, can prevent operating problems and save money.

Forward-thinking companies are realizing the benefits of integrating process control with design, says Joseph F. Boston, president, Aspen Technology Inc., Cambridge, Mass. At the American Institute of Chemical Engineers' Technochem Expo '96, May 7-9, in Houston, Boston chaired a session on integration of design and control.

Speakers at the session were:

James J. Downs, engineering associate, Eastman Chemical Co.
Vincent G. Grassi II, manager, process modeling, Air Products & Chemicals Inc.
Bjorn D. Tyreus, senior research associate, E.I. DuPont de Nemours Co.

Integration

One of the driving forces for integration of design and control is increased emphasis on product quality, according to Downs. It is no longer enough to simply meet or exceed product specifications; rather, low variability of product properties is stressed.

Eastman's design and control groups are organized to promote a high degree of interaction, says Downs. The two groups use the same computer simulation tools, for example.

In addition, Eastman's control engineers are chemical engineering graduates, which gives them a strong basis in design. Young engineers have good design skills, adds Downs, but are not adept at operability issues.

Tyreus says at DuPont the focus is on improving the productivity of existing processes. For new units, however, control and operability are considered in the "preprocess" phases.

Downs points out that there is an important trade-off between plant operability and capital cost. He suggests considering capital, operating, and what he calls variability costs when devising control strategies. Eastman measures these variability costs not in dollars, but in terms of "sigma spread."

DuPont's Tyreus describes this tradeoff as the elimination of "overdesign," which includes such equipment as buffer tanks or extra trays in a distillation column. "Plants are built snugger and snugger," says Tyreus, "making the control engineer's job more difficult."

Strategies

Downs says deciding where to set process throughput dictates much of the remaining design. If there is one piece of equipment, for example, deep in the process, that requires a steady feed rate, this is likely to be the ideal place to set the throughput rate.

Downs says Eastman starts with proportional integral derivative (PID) control, then adds layers of advanced processes on top of that, as the market dictates. Eastman uses nonlinear and model-predictive-type advanced controls.

Grassi and Tyreus described their companies' control strategies in more detail.

Air Products' control approach involves a number of elements:

Regulatory control system synthesis
Operability analysis
Advanced regulatory control design
Economic optimization
On-line process economics metrics
Predictive/inferential control
Follow-up/continuous improvement.

Grassi also outlined Air Products' integration methodology:

Determine operating objectives
Develop steady-state and dynamic models
Verify models
Synthesize controls for operability
Improve design, if necessary
Implement cost-effective, advanced regulatory control
Implement optimization when needed
Monitor solution performance vs. optimum.

DuPont uses commercial steady-state simulators, and a mixture of commercial and in-house products for dynamic simulation, says Tyreus. DuPont typically uses a five-step cycle for control scheme design:

Simulate candidate process at steady state
Use steady-state simulation as basis for dynamic simulation
Propose candidate control structures based on experience and heuristics
Modify process to improve operability
Evaluate modified process at steady state for investment economics.

Benefits

According to Grassi, Air Products can boast multimillion dollar savings as a result of integrating process design and control.

Examples of these savings include:

$2 million/year as a result of heat-integration of a distillation unit in 1983 (The benefit achieved by moving temperature measurement from Tray 4 to Tray 11 is shown in Fig. 1 [17557 bytes].)
$1 million/year from implementing a distillation control strategy in 1986
$1 million capital savings plus reduced operating costs of $1 million/year for a plant-wide control project in 1990
$0.5 million/year from operability improvements related to a 1994 project
$1 million/year from an on-line economic performance metrics project implemented in 1996.