Process Control for the Process Industries Part 2: Steady-State Characteristics

Chemical processing requires a different approach to control and automation than other types of machine and equipment control. This article focuses on the factors related to the steady-state characteristics of a chemical process.

Five aspects of chemical processing contribute to the unique nature of process control. Part 1 of this two-part article (March 2017, pp. 33–38) dealt with three factors related to the dynamic characteristics of a process: disturbances, transportation lag (or dead time), and process dynamics. Here we deal with the two factors that pertain to the steady-state characteristics of a process: economics and the multivariable nature of industrial processes.

Economics of continuous plants

Continuous processes are designed and constructed to produce commodity products in the most efficient manner possible. Margins are small, so attaining highly efficient operation is essential.

Products must be manufactured to meet specifications that are written with the objective of assuring that a product is suitable for its intended application (for example, the objective of the specifications for motor gasoline is to assure that automobiles run properly). Targets within such specifications can take two forms:

• the specification states a maximum (or minimum); product purities are usually specified in this manner
• the specification states a range of acceptable values; physical properties are often specified in this way.

Many products require multiple specifications. For example, the maximum total impurities might be stated along with a maximum for one or more specific impurities. The process must be operated so that all specifications are met.

Production personnel must operate a continuous process in the most efficient manner that makes products meeting their specifications. This is always complicated by the presence of constraints imposed by the physical equipment (the pressure at which a pressure relief device reacts), the process fluids (water freezes at 0°C and boils at 100°C or so, depending on pressure), and other factors.

With few exceptions, the economic optimum is operation as close as feasible to one or more constraints. Because all loops are subject to disturbances, the target used during operation must be safely above (or below) the constraint value. There is always a cost for this, but this provides an incentive to improve the controls so as to operate nearer to the constraining value.

The values in product specifications are treated in much the same manner as constraints. If “hot” water means a water temperature of 60°C or higher, the setpoint for the hot water temperature controller must be sufficiently above 60°C that routine disturbances do not drop the temperature below 60°C. Operating with a setpoint of, say, 65°C ensures that the product always meets spec, but at the cost of increased energy consumption.

Improving the economics translates to less-conservative targets for process operations. Assuming current operations are not significantly conservative, this is only possible when the variance about the current target can be reduced, which generally requires enhancing the performance of the process equipment and/or process controls. Such endeavors are often summarized as “narrow the variance, shift the target.”

Consider the simple example of filling a container with one liter of a beverage. The label states one liter of beverage. That does not mean one liter on the average — it means that every container contains one liter of beverage, and very few, if any, exceptions are tolerated.

So what target is specified to the filling system? No filling system is perfect. If every container must contain one liter of beverage, a target somewhat above one liter must be specified. How much above depends on the performance of the filling equipment and its controls.

Specifying a target above one liter means that product is being given away. The customer pays for one liter of beverage; if the container is filled with more than one liter, only the customer benefits. An economic incentive exists to minimize the product giveaway.

This simple example raises some interesting points. The quantity of the annual product giveaway is computed by multiplying the average excess product in each container by the number of containers produced annually, and this is multiplied by the product value to get the economic cost of the giveaway.

Those who perform such analyses must be intimately familiar with the relevant aspects of the production facility. The number used for the product value depends on the operating environment of the production facility, that is, whether it is sales-limited or production-limited:

• sales-limited — if the excess were completely eliminated, the upstream production rate could be reduced by the excess, and the annual cost of the giveaway is computed using the production cost for the product
• production-limited — if the excess can be used to increase sales, the annual cost of the giveaway is computed using the product’s net selling price.

Establishing the economic cost of the annual product giveaway quantifies the maximum possible annual return from an investment in process equipment and/or control system enhancements to reduce the giveaway. From this value and the company’s approach to evaluating capital projects, the maximum justifiable investment in a capital project to reduce the giveaway can be computed.

“Narrow the variance, shift the target” comes with a catch: The focus of control specialists is to improve control performance, but shifting targets is the responsibility of process operations personnel.

Narrow the variance. What reduction in variance can be achieved through specific process equipment upgrades, control system enhancements, etc.? These are technical questions to be answered by specialists at corporate engineering, process equipment suppliers, control system vendors, etc.

Shift the target. How much will targets be shifted because of the reduction in variance? Specifically note the question is “how much will targets be shifted,” which is different from “how much can targets be shifted.” This question is for production operations.

With increasing frequency, operators of commodity processes are looking to outside technology suppliers. Commodity processes like paper machines, power plants, refining units, etc., are somewhat tailored to the requirements of a specific facility, but usually the installation in one company has much in common with installations in other companies. There must always be a first-time application, but thereafter, the supplier can draw on experience to estimate what can be achieved.

In the abstract, casual attitudes are routinely taken toward shifting targets. But in practice, the issues are very serious. All products are purchased with some purpose in mind. The specifications for motor gasoline state the requirements that must be met so that an automobile will run properly. But the specifications are not absolute. Could targets be shifted within the stated specifications such that the final product does not perform in an acceptable manner? Fortunately, such occurrences are unlikely, especially for established products such as motor gasoline, but the possibility remains. If the problem is widespread, the specifications will be reformulated to prevent the targets from being adjusted in that manner.

Some products are sold with a guarantee of performance for a stated time. For example, asphalt roofing shingles are sold with a 10-yr (or longer) warranty. Accelerated tests provide a degree of confidence that the product will perform for the stated time period. However, such tests are not perfect — they assess some aspects well but others not so well. Suppose a product with a 10-yr guarantee begins to fail after five years. The resulting liabilities can be enormous — so much that the company may not survive. Adjusting targets in such an environment is a high-stakes action.

Economics of batch plants

Although batch processes are occasionally used for commodity products (e.g., polyvinylchloride [PVC]), most batch plants produce relatively small quantities of a variety of products. The...