9th September 2010 17:08
|
multivariable predictive control in a delayed coking plant
|
|
  |
|
Problem Statement
The fractionating column, in a delayed coking unit, is subjected to cyclic disturbances, not rejectable by conventional regulatory control techniques. Disturbances result in long periods of out-of-specifications production, reducing the unit efficiency |
Improvement Approach
A sophisticated application of advanced process control, featuring multivariable predictive control |
Benefits
Improvements of overall product quality, distillate yields, energy consumption, unit stability and efficiency.
Payback was 5 months.
|
|
|
Company Contact Information
ABB Process Solutions & Services
Tullio Grosso - ABB Process Solutions & Services - Via Hermada, 6 - 16154 - Genova - Italia
Tel.
+39 010 6073273, Fax. +39 010 6073691
E-mail
tullio.grosso@it.abb.com
www.abb.com
|
|
|
Coking Plant Layout
|
|
The coking plant of the refinery is composed of the following main units (see figure 1):
· A fractionating column (C-1)
· A column divided in two independent sections for the steam stripping of light and heavy gas oil (C-2)
· A furnace (F-1) for reaching cracking triggering temperatures. Two coke drums (V-1 and V-2).
A delayed coking plant can be described as a unit in which two separate sections live together: one continuous, concerning the cracking process, the other semicontinuous, concerning the periodic switch and decoking of coke drums. When the coke reaches the maximum level allowed in a drum, it is necessary to turn the charge coming from the furnace towards an empty twin drum and to empty suitably that just filled, by means of mechanic or hydraulic drilling. The periodical switching between the coke drums causes a series of disturbances to the fractionating column, such as considerable temperature drops, with respect to optimum values. When a drum is insulated, it is subjected, at first, to a steam stripping. |
After the cleaning, the drum must be preheated so as to replace the operating one.
During the preheating phase a fraction of vapors is deducted from the column and sent to the idle coke drum where it condenses and is drained. After these operations the coke drum is ready to start. The drum change takes place each 24-26 hours and depends on quantity and quality of residual product processed. The overall effect of these actions on the fractionating column C-1 is the reduction of the available heat with a consequent result of a sharp reduction of distillate flow rate. Conventional control system tries to compensate these reductions by closing pumparound circuits, but this action is limited by the achievement of the low flow rate constraints, and, sometimes, operators are compelled to control some variables manually. |
|
Once the Functional Specification has proved the application of advanced control technologies both feasible and profitable, the development has been carried out by following strictly a methodology which has been well established in various similar applications. The first phase dealt with plant data acquisition. The plant was stimulated by stepping manipulated variables in such a way as not to disturb normal production activity and by logging at the same time the response of all interesting process variables. Then a wider data collection (some weeks) has been carried out with the unit running normally, so as to quantify the effect of disturbances derived from drum switch on process variables of the mainfractionator. Data collection has been carried out by means of InfiView, a PC-resident software tool developed by ABB, which allows on-line and off-line display of values gathered, their file sort and export towards other SW environments through an interface based on Microsoft Windows operating systems. |
The user-friendly approach of InfiView has allowed the company staff to carry out autonomously a great number of data collections. Data files collected and suitably converted represented an input for the construction phase of impulse response models, describing dynamic response of output variables (CV) to changes of input variables (MV or DV). This phase also was completely developed by PC, through a parametric identification software called IDCOM-B SMC-Model. The amount of causal relationships expressed by the models obtained, allowed the improvement of the control diagram design, thus removing variables/models which proved themselves non-significant and inserting new ones. After the installation phase of the advanced control strategy, the controller IDCOM-B automatically managed the loops set-points, controlling feeding, distillate extraction and pumparounds flow rates, so as to control temperatures and to apply constraints due to physical operative limits of the unit. |
|
One of the main advantages to be emphasized is the drastic improvement of the process knowledge deriving from data study, collection and analysis, models creation and off-line simulation which often revealed unknown and underestimated dynamics. This aspect, joined to the revision of the basic regulation, represents a sure consolidation of the unit management, leading to the recovery of significant operative margins. The operative staff was very interested in the possibility of using the off-line simulator to analyse and verify non routine conductions. Besides, from the first tests, it was clear that the multivariable controller reduces notably operator’s engagement, especially during the critical phases of the drum change, as well as that the evaluation of the product qualities supplies a useful element measuring the unit conditions. Operative experience gained emphasized the need of managing three conditions fundamental for the success of such ambitious applications. |
·Formation of a work team between supplier and customer, who cooperate for the goal fulfilment
·Possibility of involving each unit reality, paying particular attention to relationships with operators, who must consider the innovative product a useful support and not a trouble or a source of problems
·Availability of advanced software tools, which can remove from the already difficult control challenge any complication deriving from the presence of complex hardware and software interfaces or of lacking or defective development instruments.
The last aspect was extremely important, both for the application development and for the involvement of the refinery staff.
Benefits due to increased production of distillate yields, and operation closer to costraint limits, reached the whole project payback within a period of five months.
|
|
