Pump planning when the pressure is on
Practical pump design and possible ways forward

The time available for planning is always limited, and pump engineering is no exception. It should therefore come as no surprise that when the crucial moment arrives, not all of the available tools are used and that instead engineers work with standard values and fall back on experience. It would be futile to ask for more time. The real challenge is to put the information which is generally available into a format that can be used in an efficient, purposeful manner.

Chemical systems planning normally runs in four phases: study, conceptual planning, advanced conceptual planning and detailed planning. An essential goal of the first three phases is to generate an increasingly accurate cost forecast, so that when the project is submitted for approval at the end of Phase 3, the supporting cost information is accurate to within plus or minus ten percent. Few fundamentally new issues are addressed during phases 1 – 3. Instead, a critical look is taken at assumptions that were made during the previous phase, and the issues are further defined, refined and analyzed to achieve the desired level of accuracy. This methodology also applies to pump design.

As planning becomes more detailed, an increasing number of persons from various disciplines become involved. Participation during the study phase is normally limited to project and process engineering staff. However, mechanical and materials engineers become involved during the conceptual planning phase to identify the best concept for the pump system. It is crucial that the pump is not evaluated in isolation but rather as an essential part of the pump system including peripheral devices which must be also analyzed. For an average size standard chemical pump, the investment in the total pump system is about five to six times the cost of the pump on its own. By the advanced conceptual phase at the latest, pipeline planning engineers (socket forces on hot pipes) and process control engineers (control strategy) become involved. Design planning is usually left to the detailed planning phase. As a result, some questions can be answered at a very early stage, but other details will still be unclarified when the order is placed.
Phase 1: Study phase
A reliable process simulation is normally available in the study phase or at the latest in the conceptual phase, and the required volume flow is known with a fair amount of precision very early on. Unless fundamental changes are made to the project goals (e.g. plant capacity or load factor), there will be little change to the volume flow requirements during the course of the project. This is not the case with other information relating to the pump. Very broad assumptions about the probable pump type, the expected material and the head are made during the study phase. The only object of this phase is to define the scope of the expected investment costs, which is then used to determine whether the project is likely to be a financial success.
Phase 2: Conceptual phase
The assumptions made during the study phase are revisited and questioned during this phase. The material is defined or, if that is not possible, research into the literature is initiated or lab/fields tests are commissioned. The pump type is scrutinized, and alternatives are identified and discussed. Practical experience, material properties and process requirements play a role in these deliberations. A draft layout plan is developed during this phase, which can be used to derive further pump delivery head information. The geodetic head can be determined with a reasonable degree of accuracy. Rough estimates based on practical experience with the various components must be used for dynamic head, for example 5 m for a control valve, 3 m for flow measurement, 5 m for flow losses in the pipes, 4 m for flow through a heat exchanger, etc.
Phase 3: Advanced conceptual stage
Additional information that can be used to define the pump becomes available during the advanced conceptual phase. Pipe diameter is defined. A layout plan is available, process control equipment is defined and equipment dimensioning is completed. This information can be used to identify the approximate pipe layout and the pressure losses in the inline equipment. The geodetic head can be determined quite accurately. Standard values taken from handbooks are normally used for flow losses in the pipe fittings (elbows, reductions, stop-cocks, T sections, etc.). For example, a pressure loss coefficient of z= 0.3 is used for an elbow. Equipment pressure losses (e.g. heat exchangers) are either known based on the company’s own design data, or they have to be estimated if a supplier is responsible for the design. Dividing points and overload points should also be included during the advanced conceptual phase. Due to time constraints, this is often not done, especially when small pumps are involved. Instead, generous assumptions are made regarding volume flow and head. Once again, the pump type is scrutinized, and the requirements, material properties, operating point, flow requirements, leakage requirements and availability are reviewed. Employees have access to reference libraries which are intended to enhance the decision making process. Life cycle costing would undoubtedly be very helpful at this point.
However, this is seldom done due to a lack of sufficient data. The data which has been generated using these criteria are documented by the project planning team in a process engineering data sheet. Mechanical engineers extract from this document the data that they need to define the dimensions of the pump. They also add other engineering information that will be included with RFQs and subsequent orders which are sent to suppliers.
Phase 4: Detailed planning
Normally no new information relating to volume flow and material becomes available during the detailed planning phase. The pump type is only questioned if a manufacturer injects some interesting new alternatives. The actual pipeline planning is done during this phase, and orders are placed for the process control equipment and systems. It is not until after these planning steps have been completed that exact calculations of pressure losses in the pipes and thus dynamic head can be carried out. Given the pressure to work quickly and long pump lead times, the pumps have to be ordered at this point. Recalculation of the delivery head based on the new exact data is usually not done due to time constraints. The order is based on calculations from the advanced conceptual stage. The engineers rely on the safety margin which they have built in and assume that the control system which is required in any case can be used to set the operating point. However, at the end of the day, everyone must realize that delivery head calculations were based on a certain set of assumptions. Recalculations performed on a sample basis have shown that the margin of error compared to the actual delivery head requirement is normally as high as 15 percent. Due to the safety margin, the actual delivery head is nearly always less than the value which was defined.
Delivery head always too high
Given this scenario, it is reasonable
to ask why pump selection and calculation of delivery head are so superficial. The first answer was provided above. Using an average centrifugal pump as an example, it shows how much (or rather how little) time the various departments are able to dedicate to the pump during the course of the entire project and what questions have to be answered. It is obvious that there is not enough time to conduct elaborate data investigations for each pump. So what tools are available to facilitate the work? Good process simulation tools are now available to help calculate volume flow, so this is not an issue. Determining the material to be used in a chemical pump is a very delicate matter which requires practical experience or expert advice. It would be unwise to leave this decision to system planners. As mentioned above,
we provide reference libraries to help
engineers select a suitable pump. These
libraries reflect years of company experience. They are more or less structured, and attempt to provide support in all situations and help identify the best solution. The table on page 30 shows an example for material properties. Lifecycle cost is a classic negative example. LCC equations that look very scientific are seldom accompanied by an explanation which indicates what the various terms mean and what aspects need to be considered when the data is collected. Unfortunately, the libraries do not contain a single number which can be inserted into the formula. Assuming the planners have any (company) sources available to them at all, it would take at least eight hours to assemble all of the data needed to perform a complete LCC calculation for a pump. It has been shown that knowledge of the complete set of data is often not needed in order to directly compare two solutions which are roughly equivalent from the engineering point of view.
A lack of software that is suitable for practical use
Attempts have been made for years to find ways to guide a user through knowledge libraries of this type. A decision tree is suitable for libraries that are not too
large. An IT-based expert system using
fuzzy logic could provide support with very large libraries. The main problem appears to be getting the persons who develop the methods to sit down with persons who
have the necessary plant engineering knowledge, so that they can come up with a common understanding of data reliability and validity ranges. The situation is
similar when it comes to calculating pressure losses in a pipe system, in other words calculation of dynamic head. As mentioned above, engineers normally work with standard values, although large libraries have been in existence for many years which can be used for precise calculation of pressure losses. These libraries are an open invitation to make the information available in computer programs. However, there is no recognized software on the market which does that. What you find instead in nearly every engineering department is programs for pressure loss calculations written by “hobby programmers”. When you look at these programs more closely, you discover that these programs only calculate pressure losses in straight pipes with a constant diameter using the
Colebrook formula. These programs
normally stop when things get interesting,
namely when it comes to calculating
fittings and inline equipment. The user
is normally requested to manually enter the correct pressure loss coefficients. A good program should be self-explanatory, and a user who only accesses the program twice a year should be able to use it
without making any errors. Then there
is the question of the confidence interval and guarantee. However, even this type of program, which is long overdue, should only be an interim solution. Once 3D planning for the pipeline and the inline equipment is complete, all of the information which is needed for exact pressure loss calculations is available in the CAD system database. The next logical step would be to create an interface which could be used to transfer data from the CAD database to a pressure loss calculation program. Once pipeline planning is complete, it would only take the press of a button to calculate the pressure loss for the pump.