Feature Report and second, the formula values (times, temperatures and raw material quantities). The procedural elements may differ little between products. For example, if Material 4 is not used in Product 14 but is used in all others, then a formula value of 0 is entered and the Material 4 phase does not execute when Product 14 is being produced. The same procedure may, therefore, be used for multiple products. For the purposes of this example, assume Products 1 to 18 require procedure A and Products 19–22 required Procedure B. All products, therefore, can be accommodated by building and maintaining 2 procedures and 22 formula sets. The procedure, formula and route through the plant are selected at run time. Equipment can be selected at any point during recipe execution before the unit procedure requiring that equipment executes. The plant includes shared equipment that must be acquired and released by a batch. For example if Material 4 is being delivered to B1, and R2 reaches a point in the batch where Material 4 is required, the request from R2 for Material 4 is queued. If B2 also requires Material 4 before the addition to B1 is complete, this request is also queued. The requestors are allocated on a first-in, first-out basis. Any equipment element, not necessarily an equipment module, can be declared as a shared resource. For example, only certain concurrent reactor discharge routes are valid. R1 to B1 concurrent with R2 to B2 are valid but not R1 to B2 concurrent with R2 to B1. The horizontal line between R1 and R2 is effectively a shared item of equipment and can be tagged and declared as such. If R1 is discharging to B2, the shared pipe must be acquired. If R2 reaches a discharge point to B1, the batch requests the pipe and queues until it is released by R1. With all of this in mind, it’s time to examine probably the most-obvious disadvantage of a server-based approach: system availability.
System availability System availability of a batch plant is not typically scrutinized the same way as it may be in a continuous processing facility, as interruptions to pro40
Material 2
Material 1
Pre-mixer
Material 3 Material 4
Reactor R1
Reactor R2
Blender B1
Blender B2
FIGURE 2. A typical batch scenario is shown here, as discussed in the text
duction in a batch plant are generally perceived to be easier to handle than a continuous plant, where startup time is considerably longer. Considering the potential for negative impact, though, perhaps system availability in batch plants should receive more attention. Taking into account the model above, for example, what is the estimated impact of an 8-h outage due to server failure? In the model, plant Material 2 has a short storage life and is part of the production of an upstream plant, which runs continuously. Only 6 h of storage capacity is available for Material 2. Thus, the financial impact could look like this: • Lost product with a sales value of 8 batches: $48,000 • Cost of shutting down upstream plant, which produces other products: $90,000 • Disposal of off spec production from incomplete batches: $ 5,000 That’s a total of $143,000 for a single failure. Such a failure does not only hamper the production of salable product; it results in real business disruption and thus hamstrings overall competitiveness and profitability.
CHEMICAL ENGINEERING WWW.CHE.COM MARCH 2011
Further complicating matters, that scenario doesn’t even take into account the potential non-financial impacts of safety or environmental incidents resulting from the unexpected process disruption, and the disruption to customer deliveries. By contrast, how then does a controller-based system improve on these results? A controller-based system can improve overall availability because the batches are executed on a platform that lacks a single point of failure. This approach can drastically reduce the number of process interruptions that are caused by hardware failures. But the benefits of controller-based batch can extend far beyond a reduction in failures. Moving batch execution into the controller level can actually streamline the process itself and thus improve actual business results.
Controller-based architecture In general, a single controller-based platform on which all levels of the ISA S88 procedural model execute is a far more robust solution (Figure 3). For starters, removing the server from the equation removes associated costs, security and maintenance