Designing Cooling Water Return Lines Appropriately
- Vipin Deshpande, Deputy Engineering Manager, Aker Solutions

The following article focuses on the need to analyse the cooling water network as a whole, for a synchronised pressure profile – which is one of the major activity that needs to be distinguished and requires ‘ concentrated efforts.’

Cooling water line sizing is often considered to be the simplest activity in any project, and hence it is often given less attention. However, inadequate or improper distribution of cooling water directly affects plant performance and consequently the guarantee figures.

It is worthwhile to distinguish between two different activities in cooling water network sizing:
  • Cooling water line sizing of an individual consumer as well as headers based on flow rates
  • Analysing the cooling water network as a whole, for a synchronised pressure profile
The first part may be simpler than the second, which requires more concentrated efforts. It may become an area of concern if due attention is not paid to return line sizing during the design process.

This article focuses on the second point, ie, the analysis of a cooling water network based on return line hydraulics. The article discusses the general approach of cooling water line sizing in the first section, and identifies loopholes with the help of a case study in the second. The conclusion indicates areas, where process engineers need to pay attention while sizing the return header.

A simplified scheme of a cooling tower is as shown in Figure 1.

The estimation of the pump head establishes the pressure profile of the network. This is to make sure that there will be appropriate flow distribution to the respective consumers.

For this activity, there are two ways in which one may proceed.

Option 1: Estimation of the pump head, considering the entire cooling water circuit (up to cooling tower) via different loops and then identifying the loop that requires the highest head.

Option 2: Estimation of the pump head, considering the atmospheric pressure at the consumer outlet that is located at the highest elevation.

The atmospheric pressure at the consumer outlet that is located at highest elevation in the network is achieved by having vent on return line. (See Figure1) Thus the head available due to the elevation in the return header is sufficient to cause the flow from consumer to cooling tower.

Option 2 is more frequently chosen, in view of the power saving it enables. It also ensures proper distribution without the throttling of valves, and hence is widely used. However, when using this method, appropriate sizing of the cooling water return line becomes vital.

Generally, since cooling water lines are expected to be on the pump discharge line, the velocity criteria with which these lines are sized are often considered with respect to this.

However, it shall be noted that for Option 2, the cooling water return line is exposed to atmospheric pressure by means of a vent, and will not see any pump discharge pressure. This line will be under the influence of gravity and should be sized accordingly.

In order to get more clarity on the subject, we shall evaluate the case study below.

There are two cooling water networks in a plant, one catering to the water requirement of consumers located up to the elevation of 15 metres (LP circuit), and another catering to the water requirement of consumers located between the elevations of 15 metres and 25 metres (HP circuit).

This arrangement is made in order to optimise power consumption, as the flows in both circuits are quite high.

It is always advisable to have a single header running from the plant building to the cooling tower, since minimum piping around the cooling water deck and the pipe rack will reduce civil loading of the cooling tower structure and the pipe rack.

With the above considerations, two different set of pumps of different heads are used to feed two networks, such as LP and HP. The return lines of these headers are then combined into to one header, which is laid on the pipe rack that is running up to the cooling tower.

As shown in the Figure 2, both return headers are provided with a vent at their respective highest elevations, so as to see atmospheric pressure as ‘back pressure’. Thus, it will be only the elevation difference that will cause flow from the respective consumers to the cooling tower, overcoming piping frictional pressure drop.

Reviewing this scheme drives one to analyse the influence of the HP circuit elevation on the flow of the LP circuit. A significant effect may be foreseen on flow distribution. This is because pressure at the common return point will be the same for both circuits.

At first sight, it appears that there should not be any problem with this arrangement, as both circuits are exposed to atmospheric pressure and connected to a common header at a single elevation. However, the effect of the elevation difference was not taken into account while considering the atmospheric pressure at the consumer outlet.

This is the point at which an issue with back pressure can arise.

As discussed in the beginning, the entire line sizing of the cooling water system, including headers, is calculated based on the pump discharge pressure. This calls for the velocity in the pipeline to be in the range of 1.5 – 2 m/s, as per the recommended criteria.

However, in reality, the return headers are expected to run under the influence of gravity.

Secondly, it shall be further noted that the atmospheric pressure that is considered as ‘back pressure’ is at two different elevations. Hence, although both headers are at atmospheric pressure, the respective elevation matters, in terms of the liquid rise in the respective headers, and thereby the back pressure on the LP circuit.

In order to avoid this effect, it is required to size these headers for gravity flow, considering a lower velocity.

Furthermore, it is required to size these headers as self-venting lines that are running vertically, so as to release any vapors present therein.

This can be ensured by getting a dimensionless Froude number less than 0.3. This can be estimated as below.

Froude Number = V2/ Dg < 0.3
V = Velocity in pipe line, m/s
D = Diameter of pipe, m
g = Gravitational Constant, 9.81 m/s2
Ref: Perry’s Chemical Engineer’s Handbook

With the above consideration, the return header size will be sufficient enough to remove the effect of elevation difference. Thus the vertical portion of the header will have a liquid leg filled to the height of the cooling tower, plus height equivalent to the frictional pressure drop of the return header.

Hence, each circuit will see the same pressure profile across the entire length of the return header. Accordingly, the HP circuit will not influence the LP circuit or the return header. The cooling water network can run then smoothly without any need for throttling valves for flow adjustment.

Although a cooling tower network works under pump pressure, a return node shall be identified from where gravity flow takes place. Accordingly, the return header shall be sized considering the guiding parameter of the Froude number, which shall be maintained below 0.3.

Secondly, it is equally important to analyse whether the elevation difference between the highest consumers of cooling water and the cooling tower deck is sufficient to drive the water under influence of gravity. If the elevation difference is not sufficient, the pump head calculation shall be done up to the cooling tower deck, and the entire circuit will operate under pump pressure. However, in that case, due care shall be taken during start-up to ensure that no air pocket exists in the loop.

Perry’s Chemical Engineer’s Handbook