Batch flow pipelines

In the batch flow system, the pipeline connects two locations where a set of products is produced and consumed. The subset of products that are used at both locations is denoted by S. These products have to be balanced between both locations by pipeline transports. 

In the literature, basically two directions can be distinguished. A concise comparison of the problem features tackled in both directions is contained in Table 3.5. 

Both directions commonly assume that the pipeUne remains completely filled over the total time horizon. The pipeUne is fed by only one tank at a time and can only feed one tank at a time. Conversely, a tank can either be filled or emptied at a time. It is assumed that for each product a set of tanks is available and it has to be determined when to use which tank for pumping. 

Let w denote the production and consumption rate for a certain product and p denote the pump capacity at which the pipeline is operated. Furthermore, let T denote the cycle time between two successive batches of the considered product and r denote the time necessary to transport a batch of a product from provider to consumer location. The 

Since production and consumption rates are equal, the total stock in the system is constant at the level L T) = L + r lo = ui + T and depends only on the cycle 

---

A more complicated problem is met in the multi-product case where batches of products have to be routed through the pipeline (network) whereby a batch may also be split into sub-batches at pipeline forks. In the oases of multiple sources, additionally a coordinated pumping planning is required under consideration of the pumping capacity, the due dates of the batches, and the segment-wise flow balances. 

As the Uterature on pipeUne scheduling has its roots in the petrochemical industry, a straight-forward extension from batch flow pipeUne systems to batch split systems is natural. Here, typically a single source has to serve a set of distinct sinks with a set of (petro-)chemicals. Typically, the source represents a refinery and the sinks represent local fuel depots from which the local customer demand is served. Conversely, a harbour may also serve as a supplier of different types of raw materials for a set of refineries. These sinks are connected to the source via a serial pipeline system. At the source and the sinks a set of storage tanks is available for each product. Common assumptions for the proposed models in this branch of literature are summarized as follows ... 

Metal ore bodies may contain several different minerals that are separated into individual concentrates. These may be slurry-transported in the same pipeline by pumping them in separate batches, each separated by a slug of water to prevent contamination [607]. Such batching also allows pipeline flow to be maintained when the mine or separation site temporarily runs out of ore. 

Emulsion Pipeline Operations. Prediction of pipeline pressure gradients is required for operation of any pipeline system. Pressure gradients for a transport emulsion flowing in commercial-size pipelines may be estimated via standard techniques because chemically stabilized emulsions exhibit rheological behavior that is nearly Newtonian. The emulsion viscosity must be known to implement these methods. The best way to determine emulsion viscosity for an application is to prepare an emulsion batch conforming to planned specifications and directly measure the pipe viscosity in a pipe loop of at least 1-in. inside diameter. Care must be taken to use the same brine composition, surfactant concentration, droplet size distribution, brine-crude-oil ratio, and temperature as are expected in the field application. In practice, a pilot-plant run may not be feasible, or there may be some disparity between pipe-loop test conditions and anticipated commercial pipeline conditions. In these cases, adjustments may be applied to the best available viscosity data using adjustment factors described later to compensate for disparities in operating parameters between the measurement conditions and the pipeline conditions. 

For example, cesium-137 can be used to monitor the flow of oil in a pipeline. In many cases, more than one oil company may use the same pipeline. How does a receiving station know whose oil is coming through the pipeline One way to solve that problem is to add a little cesium-137 when a new batch of oil is being sent. The cesium-137 gives off radiation. That radiation can be detected easily by holding a detector at the end of the pipeline. When the detector shows the presence of radiation, a new batch of oil has arrived. 

The considered plant (Fig.l) involves 13 areas (nodes), including 4 refineries (nodes Nl, N3, N9, and Nil) and 2 harbours (NIO and N13), which receive or send products through 7 distribution terminals. In addition, it includes 29 multiproduct pipelines with particular volumes e.g. pipe 1 has more than 42000 m ). Nodes are connected through pipes (e.g. pipes 3, 4, and 5 connect nodes N2 and N3). A product can take many hours to reach its final destination. A batch can remain in a pipe imtil another one pushes it. Many pipes can have their flow direction inverted due to operational procedines e.g. pipes 5, 7, and 15). Each product has to be stored in a specific tankfarm within a node. More than 14 oil derivatives can be transported in this network. Adjacent products can share an imdesirable interface, e.g. alcohol pumped just after diesel. In this case, it is necessary to piunp another product between them e.g. gasoline). Typical transfer tasks can involve pumping a batch through many areas. For instance, a batch can be pumped from node N3 to N7 through nodes N2, N5, and N8. In that case, the batch uses pipes 4, 8, 12, and 14. 

A type of continuous reactor with performance similar to a batch reactor is the plug-flow reactor, a tubular or pipeline reactor with continuous feed at one end and product removal at the other end. The conversion is a function of the residence time, which depends on the flow rate and the reactor volume. The data for plug-flow reactors are analyzed in the same way as for batch reactors. The conversion is compared with that predicted from an integrated form of an assumed rate expression. A trial-and-error procedure may be needed to determine the appropriate rate equations. 

Material flows along multi-product pipelines are far more complicated to plan. The most prominent field of application for multi-product pipelines is the distribution of refinery products among a network of distribution terminals. The operating principle is essentially the same as for single product pipelines. However, as multiple materials pass the pipeline sequentially, the different materials have to be separated in some sense. Basically there are two options to separate batches of different materials... 

Naturally, the quality of these decomposition approaches is considerably affected by the neglected interdependency of the decomposed decisions. Therefore, monolithic, time and volume-continuous MILP formulations are proposed recently to integrate these decisions. These models are straight-forward adaptations of one-to-many approaches. The basic setting is adapted from the one-to-many case (see the previous paragraph), i.e. a serially operated main pipeline is assumed with a refinery at its head and depots along this pipeline. One key alteration is that depots now can receive and inject batches in the pipeline. Most other assumptions are inherited from the one-to-many systems. In particular, the product flow is stiU unidirectional and only one location at a time can inject a batch in the pipeline. ... 

The material flow system is complex because the mixers and machines have to be changed in accordance with the product type and batch size. Equipment also has to be reserved for specific product groups with special shades or chemical properties. Only in this way can the effective cleaning of equipment and pipelines during batch changes be guaranteed. 

A key operating parameter is pressure drop across the pipeline, for which a knowledge of the emulsion s rheological properties is needed. Although use may be made of batch rheological measurements (Section 6.2), the final data for scale-up are usually obtained from viscosities determined in a pilot plant, involving large batches of the prospective emulsion that are made to flow in a pipeline loop. 

The size of a filter, in terms of fabric surface area required, is specified approximately in terms of the volumetric flow rate of gas to be handled. Filtration plant is designed for steady state conditions and so if there are likely to be surges and transient effects in performance due to batch conveying or pipeline purging, for example, these effects must be taken into account. Filter fabrics will not last forever, and it is important that maintenance schedules are kept. 

The flow rate of solids is measured by means of load cells beneath a receiving hopper at the end of the pipeline. The accuracy of the weighing system places finite limits on how short a test run can be used for data gathering, because it is necessary to resolve relatively small differences in large values. For example, under lean phase conveying conditions the flow rate of solids might be as low as 0.25 kg/s, whereas a total batch of solids may be as large as 1000 kg with a load cell accuracy of nominally 1 part in 10,000 for this set-up (typical of simple... 

Thus no products should be made, no components ordered, until there is a downstream requirement. Essentially JIT is a pull concept, where demand at the end of the pipeline pulls products towards the market and behind those products the flow of components is also determined by that same demand. This contrasts with the traditional push system where products are manufactured or assembled in batches in anticipation of demand and are positioned in the supply chain as buffers between the various functions and entities (see Figure 5.5). 

---