PORT STUDY SOLVES REFINERY SHIPPING BOTTLENECK

A.J. Trenchard, Vaughn Bull, M.J. King
KBC Process Technology
Southampton, U.K.

Congestion at product shipping facilities at Sweden's 200,000 b/d Scanraff refinery at Lysekil prompted a study of the facilities there. It led to recommendations that are expected to cut berth occupancy and demurrage costs.

Product shipping facilities common to many refineries are often bottlenecked. This can lead to unnecessarily high demurrage costs as well as other performance-related problems. One of the major obstacles to optimizing the efficiency of the shipping operation is that the task usually falls into different subject areas, from management organization and information technology through to detailed process design and control. Understanding and solving the multifaceted problems involved, therefore, demands both a wide breadth and depth of expertise which is often difficult to assemble. As a result many shipping facilities do not operate optimally.

The Scanraff study was completed in 1989 and involved visits to the refinery and the owners' headquarters by a team of KBC consultants to review the shipping and related issues with the key parties. The resulting recommendations have since been implemented by the refinery.

LOADING DIFFICULTIES

The principal reason for the investigation at Scanraff was the unusually high level of demurrage claims being made against the owners; however, the refinery performance was also penalized in the following ways:

once ship owners became aware of congestion at the port there was a general reluctance to charter vessels to the port. This meant additional cost penalties in terms of higher freight charges.
The immobility of the port restricted potential trading opportunities.
Shipping delays occasionally changed the refinery operation from the optimum in order to meet shipping movements.

Management had already identified that the cause of its problems stemmed from the high average berth occupancy of 60%, causing ships to wait at anchorage prior to loading. This, in turn, was mainly due to the product movement flexibility requested by the refinery owners in terms of the wide range of products and small parcel sizes.

In an attempt to alleviate the problem, the construction of a new loading jetty was considered; however, the capital cost of $7 million could not be justified in terms of payback alone. Furthermore, the benefits of increasing the number of loading berths were unclear, given that the loading operations were frequently limited by offsite constraints. The refinery management, therefore, requested a team of KBC consultants to investigate the alternative options.

Five main areas were targeted for review, namely communication, planning, jetty operation, process design, and process control. These issues are discussed in more detail in the following sections.

COMMUNICATION

The transfer of shipping and product inventory information between the refinery and its owners was found to be one of the weakest links in the shipping operation. All three owners used clear and efficient lines of communication to transfer essential operational data to and from the refinery; however, the dissemination of information key to the optimization of the shipping operation was less well-supported by all the parties.

For example, the demurrage charges, incurred because of shipments from the port, were not readily available to the refinery planners. As a consequence, the refinery management did not have a complete picture of the performance of the shipping facilities vs. the operational demands.

Similarly, the limited ship performance data (listing those vessels which constrained the loading operation by restricting loading or deballasting rates) was not communicated back to the owners' shipping departments. The owners therefore did not have the benefit of past performance to assess the best ships to charter. Jetty statistics had once been sent to the owners, but the procedure had been discontinued without objection from the owners, following the lack of feedback.

This may have been because the information was too detailed, did not identify the financial implications of the jetty operation, and was therefore of limited use to the owners' shipping departments.

The owners also appeared to lack scheduling information about each others product shipments. While they were to some extent in competition to sell their products, it became apparent that it would be in their mutual interest to consider coordinating some of their shipments, for example by jointly lifting products which would otherwise be shipped in small parcel sizes.

After discussion with the refinery and owners, it was suggested that the owners could enhance the structure of their interfaces with the refinery, to not only retain the existing clear lines of communication, but to also enable wider dialogue where necessary. More specifically, it was agreed that there was a need to improve liaison between the owners, charterers, traders, and the refinery planners.

In addition, all parties were encouraged to transfer information which would improve the optimization of the shipping operations. For example, the owners could endeavor to detail the demurrage charges on a ship-by-ship basis. Similarly, the owners could be regularly informed of the past and predicted jetty utilization. As a future objective, all the parties involved could be motivated to use and contribute to an enhanced refinery information system, leading to a significant integration of effort.

PLANNING

The planning cycle operated by the owners and the refinery was run to a regular monthly schedule. This was principally because the product allocation calculations were performed once a month. The owners, therefore, had the most reliable information on their actual stocks once a month. As a consequence, the selling activities of the traders were uneven, tending to peak after the calculation of stocks, which then caused a peak in the product shipments later in the month.

Both the owners and the refinery recognized that this operation bottlenecked the shipping facilities for part of the month.

KBC suggested that a more flexible and rolling planning cycle was needed which should allow the owners to track their stocks of material more closely.

Using the existing procedures, this would clearly put greater demands on the refinery planning team; however, the traders would benefit from the timely information.

For example, the traders would be able to negotiate the sales of product to third parties based on the market requirements rather than being constrained by historic planning information. More importantly, the owners would avoid any tendency to directly compete against each other to sell their third-party volumes immediately following the monthly notification of stocks

The planning issue was further complicated by the differences in planning tools used by the refinery and the owners. All four parties used linear program models of the refinery; however, there were certain differences between the models and the data used within them. Furthermore, the models were driven slightly differently by the owners and the refinery.

For example, the refinery planners would tend to set higher production targets in the first half of the month as a safeguard against possible shortfalls later in the month, whereas the owners generally assumed even production throughout the month. The most notable consequence of the divergence in planning tools was the need to ship high levels of product at short notice, adding to the congestion at the port.

After discussion with all the parties, it was agreed that the solution to the problem was to ensure that the owners and the refinery used the same models and data to forecast production and to facilitate the communication of operational constraints and objectives to and from the owners' planning departments.

In the short term the situation would be improved by increasing the coordination between the planning departments.

In the longer term, the consistency of planning would benefit most from the centralization of the planning models at the refinery. These would then be accessed by the owners via computer links. Adopting this approach would ease model maintenance, remove the duplication of information, and enable the most reliable and timely process data to be used in updating the models.

While the improvements in planning coordination should alone justify the centralization of the production planning models, one of the most important additional benefits would be the ease with which the same information could be integrated into the other refinery planning tools.

For example, the scope of the jetty occupancy planning model could be expanded to take account of predicted refinery production and hence product levels.

This information, in conjunction with historical ship loading performance data (where available), could then be used to highlight future jetty congestion resulting from the shipping plan. Furthermore, the same information could be used by the owners to simulate various "what if" scenarios, for example, to determine the benefit of chartering different types of vessels with higher loading or deballasting rates.

From the refinery standpoint, if the owners used the centralized planning models there would be an inherent improvement in information exchange between the refinery and the owners. The consistent plans would also better enable credible operational targets to be set.

JETTY OPERATIONS

Unlike many refineries, Scanraff had already installed a jetty-monitoring system. Excellent though this was, it had not been integrated with the refinery information system. While certain shipping information was logged on computer and paper, there had been limited analysis of the data and little feedback into the planning models. This operational strategy contrasted with other refineries of a similar technology level, where jetty monitoring systems had been integrated into the scheduling information system.

The benefits of such systems fall into two main categories. First, the quality of the product shipment plans depend heavily on the accuracy and detail of the information used to formulate them. For example, if the plans assume unrealistically short product loading times, then the schedules will never be met. Conversely, if an unnecessarily high margin of error is assumed, then the jetty performance will be nonoptimal. Such data can only be obtained by closely monitoring the current performance of the operation in question.

The second main area of benefit lies in analyzing the performance data to identify the strengths and weaknesses of the jetty operations facilities. For example, in preparation for the shipping study at Scanraff, the refinery management and operational staff prepared a detailed review of the product shipping operations over a 2-month period. From this information the KBC consultants were quickly able to identify the specific areas for further investigation.

As an example, the jetty operations of berthing, connection, disconnection, and deberthing were found to be efficient; however, the procedures for certifying the loaded product qualities frequently caused long time delays.

Unfortunately, the preparation of the detailed shipping review proved to be quite time-consuming, and therefore Scanraff management was reluctant to perform the same task on a monthly basis. This was because the necessary information had to be gathered from a number of different sources, from paper logs to computer printouts, and then cross related.

Clearly, there was a need for further investigation into the integration of the refinery information and the computer-based tools available to manipulate these data.

QUALITY TESTING

As stated earlier, a significant number of shipping delays were attributed to the process of certifying the quality of the loaded products prior to the departure of the vessels.

The actual procedure for assessing the quality of the products depended both on the type of product and its destination. For example, most of the product liftings by the owners did not require independent inspection, except for jet fuel. The quality of the loaded product was therefore estimated by testing a composite of spot samples (taken from the shore side installations), blended according to loading rates and sample intervals.

However, the procedure was somewhat different when independent inspection was required for third-party sales. In this case, the inspector generated a composite from samples taken from the loaded ships tankage. This was then tested either by the refinery or an independent laboratory.

Three main problems were identified with this operation. First, the differences between the refinery and inspectors' sampling techniques introduced discrepancies in the quality assessments which were sometimes difficult to reconcile. As a result, the loaded ships remained berthed while the differences were resolved.

Second, the procedure for sampling the loaded ships' tanks and generating a composite could only be performed once the loading operation had been completed. This again delayed the early departure of the vessels. Finally, while the efficiency of the refinery laboratory was recognized as being comparable with pacesetter refinery levels, the high testing workload imposed because of the blending operations sometimes constrained the turnaround time for the final quality analyses. During this time, the loaded vessels remained at their berths awaiting the final quality documentation.

In response to these difficulties, KBC suggested investment and noninvestment solutions. The problem of differences in sampling techniques could be resolved if an automatic in-line sampling system (agreeable to both the refinery and the inspectors) was installed. The composites generated by this route could then be tested by the refinery laboratory, and if necessary an independent laboratory with a minimum of discrepancy.

Similarly, the high laboratory testing workload could be reduced if greater use was made of on-line analyzers for limiting blending quality specifications, such as viscosity, density, and sulfur content. This issue is discussed later with reference to the control of the blending operations.

The third suggestion involved no investment, Given that the loaded vessels tied up the jetty berths while waiting for the final quality documentation, some of the ships could depart with a basic bill of lading specifying the volume and density of loaded products. The detailed quality information could then be faxed to the ship following the analyses.

While it would be unlikely that such arrangements could be negotiated for third-party sales, the owners could be encouraged to adopt the procedure especially when the port was congested.

PROCESS DESIGN

The offsites at Scanraff were well designed to match the operational requirements, therefore no major process design problems were identified. However, many smaller design improvements were suggested to debottleneck certain aspects of the loading operation. Following are two examples that illustrate the type of design issues that were discussed.

JETTY DESIGN

A schematic diagram of the product loading jetty at Scanraff is shown in Fig. 1. The jetty contains four berths, three for the main products and a fourth for LPG only. Berths 1 and 4 are designed for small coastal vessels up to about 8,000 and 5,000 tons, respectively. Berth 1 has three loading arms which are connected to a total of seven main products: two gasoline, one naphtha, two gas oil, and two fuel oil. The nominal capacity of each arm is 750 cu m/hr.

Berths 2 and 3 are for larger vessels and have identical layouts. There are six arms on each berth, and a total of eight products can be loaded; the seven listed for Berth 1, plus jet fuel. The nominal capacity of each arm is 1,500 cu m/hr. Berth 4 is for LPG only.

From the jetty-utilization statistics prepared by Scanraff, it could be seen that the LPG loading berth occupancy was low. Further analysis of the data revealed that 12% of all gasoline shipments, 8% of all gas oil cargoes, and 2% of mixed cargoes were less than 4,000 cu m. Therefore, in principle, 22% of all shipments could be loaded from the LPG berth.

The cost of installing loading capability for gasoline/naphtha and gas oil at Berth 4 was estimated at $1.6 million. With the increased jetty capacity, it was also estimated that the average occupancy would reduce to around 55%, yielding a payback period of approximately 2 years.

BLENDER DESIGN

While Scanraff possessed some blended product tankage, a high proportion of the product was in-line blended directly to waiting vessels.

In the latter mode, some delays were encountered because of maximum flow constraints within the blenders. Therefore, these constraints were investigated to determine if the throughput could be increased.

The analysis of the blender constraints was performed on a spreadsheet using typical blend recipes and information about known pumping and metering constraints.

For each blender, the component flows were calculated from the total flow and the recipe information. As the total flow was increased, all the input and output flowrates were checked to determine if a pumping or metering constraint had been met. Providing that the constraint could be overcome by design or operational changes, the procedure was then repeated with successive constraints until a practical upper limit was reached.

The results of the analysis were then tabulated as a list of maximum flow constraints and the design/operational changes required to overcome each constraint.

In many cases, the blender throughputs were limited by the component flow meters. After contacting the meter manufacturer it was found that some of the maximum flowrates quoted by Scanraff were below the normal maximum design flowrate.

The upper flow limit was in reality imposed by the flow instrumentation and could be adjusted at minor cost. Furthermore, many of the turbine meters could be operated at 25% above the normal maximum design flowrate for 20% of the duty cycle, provided that the pressure drop over the meter did not exceed 96 kPa.

By simply recalibrating the blender instrumentation, quite significant maximum flow improvements could be achieved; in the case of the gasoline blender an 85% increase in throughput. With further investment in larger pump impellers and flow meters, the potential increases in throughput were even more significant.

PROCESS CONTROL

The general level of offsites automation at Scanraff was very good; however, there was scope for improvement in the control of the blenders. At the time of the study the three blenders (gasoline, gas oil, and fuel oil) were controlled using "Jiskoot" blending controls.

These included a total flow controller (for each blender) and a slave controller for each blending component. The operators specified the total flow to the master and the percentage of each component to the relevant slave.

The basic blend recipes were calculated by the planners using a program on the refinery's Data General computer. This took into account the most recent analyses of the blending component qualities and the scheduling constraints imposed by the planners.

During the blending operation the operators took spot samples of the product which were analyzed by the laboratory to determine if the quality specifications were being met. The analysis results were then entered into the blending program via console in the control room, causing the recipe to be reevaluated.

Given the lack of on-line analyzers for the limiting of blended quality specifications, the product giveaway was very small. This was principally due to the skill of the operators and the laboratory technicians. However, this blending strategy did incur quite a high testing workload for the laboratory because of all the mid-blend product analyses. Furthermore, it was not uncommon for the total throughput of a blender to be reduced significantly while the results of an analysis were awaited. This in turn delayed the departure of some vessels.

The preferred solution to the problem involved the installation of on-line analyzers for the limiting quality specifications; however, these were difficult to justify based on any reductions in jetty occupancy because the effect on the final demurrage charges was unclear, especially if some the other KBC recommendations were implemented. The product quality giveaway figures for the previous year were therefore analyzed in order to predict a giveaway cost for the following year.

Using cost figures supplied by the planning departments, the total giveaway for 1989 was estimated as $1.5 million for the gasoline, gas oil, and fuel oil products. Therefore KBC recommended the installation of motor octane number and vapor pressure analyzers on the gasoline blender; cloud point and sulfur analyzers on the gas oil blender; and density, viscosity, and sulfur analyzers on the fuel oil blender.

Fig. 2 shows how the analyzers could be used to improve the control of the blending operation. Ideally, the control computer should have access to the total flow, all the component flows, and the analyzer measurements. The existing Scanraff blending model could then be used to predict the initial recipe based on the off-line analyzed component qualities, which would then set the setpoints of the individual flow controllers. After some delay, the analyzer would start to provide feedback on the quality of the blended product, which could then be used to recalibrate the blending model, typically at a frequency of once every 8-10 minutes.

Into this structure it would also be possible to incorporate economic optimization based on netback costs of the components and pacing (total flow maximization) typically to within 2% of the hydraulic and metering constraints.

Following the review of the product shipping operations at Scanraff, KBC prepared a detailed report which contained 45 main conclusions and 50 recommendations, which were subsequently implemented.

The main conclusions and determinations of the study and resulting broad recommendations are summarized in the accompanying box.

It was estimated that the implementation of these recommendations would allow operable berth occupancy to be cut to a target value of 55% for the foreseeable future and reduce demurrage costs in the order of $1 million/year.