Optimizing Petroleum Refinery Cooling Water System

Paresh S. Parekh, Unicade Inc.

John A. Latimer, Equilon Enterprises LLC


A 145,000 Bbl/day refinery identified cooling water(CW) circulation rates and supply temperature as the major bottlenecks to sustain crude throughput during the summer months. These CW limitations led to processing slowdowns or increased flaring of light hydrocarbons, incurring several million dollars worth of lost opportunity. The refinery has two closed loop cooling tower systems, cooling tower 1 and 2, with design flows of 74,000 gpm and 31,000 gpm respectively.

The objectives were to investigate fluid flow, heat transfer, and optimal asset utilization to obtain the best "bang" for the capital improvement dollars. Both CW systems consist of a cooling tower, CW supply pumps, CW piping distribution and CW process heat exchangers. The evaluation included creation of a complex piping network fluid flow computer model of the CW systems that were tied to an Excel based heat exchanger model which predicted interactive impact of physical and process changes of CW system on the refinery operations. The project analysis included technical feasibility, economic analysis, prioritizing alternative solutions and making recommendations. The article will address three separate areas.

  1. Problem definition, goal settings, and base case model development of two cooling water systems utilizing a PC based Flow Network Analysis (FNA) software and an Excel based heat exchanger process model.
  2. Alternative case studies with "what-if" computer simulation, new project identifications, and cost/benefit analysis.
  3. Project implementation, impact evaluation, and lessons learned.

Defining the Pre-Retrofit Case

The petroleum refinery was built in 1958 with a capacity of 44,000 barrel per stream day (bpsd). Over the period of time the existing process units were debottlenecked, expanded and new process units were added. The current operating capacity of the refinery is 145,000 bpsd and it has one crude unit, a fluid catalytic cracking unit, a cat-poly unit, two catalytic reforming units, two hydrotreating units, two alkylation units, a delayed coking unit, and a sulfur recovery unit. Two cooling towers serve the refinery processing units. The cooling tower no.1 (CT1) had a measured flow of 74,295 gpm with four pumps in operation and the cooling tower no. 2 (CT2) had a measured flow of 31,744 gpm with two pumps in operation. The cooling tower 1 was expanded in 1989 by adding one more cell and upgrading the circulation pump motors to 1000 hp. Both cooling towers independently serve the process units, and interconnection between the towers and their pipe headers is not provided. Both cooling tower distribution systems operated wide open and have no flow control. The goal of the base case was to establish a piping network model of the existing system operation. Once the model achieved the accurate mapping to the field verified data, then the "what if" case studies were conducted to debottleneck the cooling water system.

Problem Definition

Historical performance documented that both cooling water systems were not providing the desired cooling during the summer months. The cooling tower 1 normally operated with three cooling tower pumps. During the summer months, a fourth pump was brought on-line in an attempt to provide additional cooling. An increase in the supply temperature from cooling tower 2 was consistently the constraint in the CT2 system. The results of these limitations were the flaring of light hydrocarbons, reducing the throughput of various processing units and the use of external water sprays on some critical shell and tube heat exchangers to keep the process temperature down. The water run-off from the external water sprays created its own hazards, and increased waste water treatment load. The lost opportunity costs were estimated at $3-5 million per year and provided substantial justification to evaluate ways to improve the system.

Project Goals

The overall objective was to be able to sustain unit capacities during the summer months without flaring and without using external water sprays. Intermediate milestones included analysis of the pre-retrofit cooling water systems (includes cooling tower, cooling tower water circulating pumps, piping, process-to-water heat exchangers), identification of existing system limitations and screening of the major alternatives. To be successful, the project also needed to develop low cost creative solutions which maximized the existing cooling water supply and distribution system investment.


The base case (pre-retrofit) computer model development process was broken down into three separate areas: Data Collection, Model Development, and Testing / Verification. The Figure 1 outlines the approach.

Figure 1. Base Case Model Development Process

The data collection step included collecting process and design information such as equipment datasheets, cooling water distribution Process and Instrumentation Diagrams, piping layout drawings (above ground and under ground), operating procedures, etc. Field measurements of cooling water flows, cooling tower inlet and outlet temperature, heat exchangers' water outlet temperature, and cooling tower circulation pump motor power draws were taken for a total system analysis. The field measured data was then used to establish a "measured value" base operating case for each cooling water system and served as the basis for 'tuning' the FNA computer models. The goal was to create a base case flow network model that was field tested and verified.

The field flow measurements on the cooling tower circulation pumps and exchangers were taken with a "clamp-on" type ultrasonic model MST-P flowmeter manufactured by Polysonics Inc. of Houston, Texas. The flowmeter readings were reasonably accurate for the main cooling water flows and provided excellent cross checks with the power draws. However, the accuracy was poor in most cases when the readings were taken at individual heat exchangers. As a result, the hydraulic modeling was the primary source for generating the flow data for the individual exchangers. The validation checks used for the individual exchangers were the calculated cooling water heat loads and comparison of these with estimated heat loads from the process side of the exchangers.

A separate base case model for each cooling tower system was developed. The cooling tower 2 serviced four process units and cooling tower 1 provided water for the eight other refinery units. The cooling tower 1 was designed to remove 3 times more heating load than the cooling tower 2. In the subsequent presentation, we will focus on the cooling tower 2 model development and its analysis.

A base case block flow diagram of cooling tower 2 water distribution is presented as Figure 2. As you can see, the Alkylation unit is closest to the CT2 with a 30" water supply (& return) header. An analysis clearly indicated that due to a close proximity of the Alkylation unit to the cooling tower, the process unit received higher percentage of the water flow and removed lower percentage of heat load than the design.

Figure 2. A Block Diagram of Cooling Tower 2 Water Supply System (Base Case)

The flow network analysis was conducted by creating first a flow diagram of all nodes and pipes, and entering physical parameters of each pipe segment (length, type, internal diameter, elevation, fittings, valves, etc.) as well as pump curves & heat exchanger data into the modeling software program. The output results were analyzed and the node diagram was corrected until converged results agreed with the field measurements. Figure 3 is a representative node/pipe diagram for an entire system and Figure 4 is an example of one of the process units.

Figure 3. Overview Pipe Network Model Served by Cooling Tower 2

Figure 4. Detailed Pipe Network Model for the Reformer

Flow Network Model

The FNA software program is designed to calculate the flow and pressure in a "network" of pipes and piping systems. The network diagrams (Figure 3 & 4) were translated numerically and inputted into the software for analysis. The diagrams consist of a set of "pipes" and "nodes." The "pipes" duplicate the pipe sections and equipment in which the liquid is flowing and the "nodes" duplicate the connections in the system. The FNA model uses the numerical inputs to calculate a flow (in each pipe) and a pressure (in each node) by use of conventional hydraulic equations in a set of simultaneous equations. The FNA model outputs also included other data such as the Reynolds Numbers, friction factors, head loss/increase, and the accuracy of the model convergence. In both base line FNA models, the active "specified" nodes were set at atmospheric pressure at the cooling tower and the total flow of the system was permitted to balance within the model. There were no artificial constraints on the system. The results of computer simulation (e.g.: flow and pressure) were verified with the field readings. Power measurements at the pump motors allowed us to back-calculate pump flow rates and these agreed well with the final base line model. A summary comparison of flow and circulation pump discharge head is presented in Table 1 and heat duty in Table 2 for both systems.

Cooling Tower #1
Cooling Tower #2
3 Pump Operation
4 Pump Operation
2 Pump Operation
Base Case

Base Case

Base Case

Total Flow, gpm
Disch. Press., psig

Table 1. Flow & Pressure Comparison: Cooling Tower Base Case to Field Measurements

Cooling Tower #1
Cooling Tower #2

Original / Prior Exp.
Base Case

Base Case

Heat Duty, MMBTU/hr
1,100 / 763
Water Flow, gpm
50,000 / 61,000

Table 2. Cooling Tower Heat Duty Comparison: Design to Base Case FNA Model

Figure 5 & 6 below present a comparison of cooling water flow & heat duty distribution between the design case of all cooling water users in process units served by cooling tower 2 to the cooling water flow & heat duty distribution predicted by the FNA/Excel model.

Figure 5. Cooling Tower 2 Flow Distribution to Process Units: Design Case to Base Case

FNA Model

Figure 6. Cooling Tower 2 Heat Duty Distribution to Process Units: Design Case to Base Case

FNA/Excel Model

Heat Transfer Model

The flow network analysis results from each case study were integrated with a heat transfer impact analysis model for heat exchangers. Each heat exchanger was analyzed using its design data, current operating performance, and future impact on heat transfer as cooling water flow changed. It was important to evaluate the impact of water flow changes on heat exchanger tube water velocity (v), overall heat transfer coefficient (U), exchanger heat duty (Q), and overall impact on process performance. Twenty-eight process-to-water heat exchangers were identified as "problem" or "key" heat exchangers because of their poor performance due to insufficient surface area, low cooling water flow rate, higher cooling water supply temperature, excessive fouling and other process issues. The main focus was performance impact on 'key' heat exchangers that had been identified as the process bottlenecks. The impact of water flow changes on all the heat exchangers in each process plant were compiled for evaluation.

Selected results from the FNA model were transferred by an Excel macro to an Excel spreadsheet. The macro also calculated and prepared a table that summarized cooling water flow and heat duty for the specific case study and compared its results with the design case and the base case model. An example of the output is shown in Table 3 in the "what If" case study section.

"What If" Case Studies

Two independent base case flow network models, one for each cooling tower, were developed and integrated with the heat exchanger evaluation model. The field verified models were used to determine the effect of mechanical changes on the cooling water system, to provide the basis for further engineering analysis as well as for economic justifications for potential improvements. The improvements in the heat exchanger system were further qualitatively identified with other benefits such as product quality improvement, reduced flare, reduced waste water treatment, energy savings, increased throughput, lower run-down temperature and off loading of the cooling towers. In addition, we addressed the issue of net positive suction head and cavitation problem at recirulation pumps at cooling tower 1, particularly during the summer months when water suction temperature was higher and a fourth pump was brought on-line to meet the additional cooling demand.

A partial summary from a typical "what if" case study is presented in Table 3 below. Numerous case studies were conducted leading to a final package that included an overall solution that could be implemented in smaller discreet packages.

Table 3. Cooling Tower 2 Flow and Heat Duty Analysis

Findings and Recommendations

The analysis of the base case and "what if" cases revealed several significant factors about both the cooling water systems. The thorough understanding of the system performance helped us develop a series of improvements. The significant findings on cooling tower 1 and 2 are presented below.

Cooling Tower 1 Findings

Cooling Tower 2 Findings


The "what if" case studies of FNA models resulted in several recommended projects for the refinery Each proposed project was evaluated with a problem statement for the particular unit, an impact evaluation on current operation, solutions investigated, process calculations, equipment specifications, order-of-magnitude cost estimate, and recommendations with benefits specific to each project. The proposed projects included the following.

No modification to the cooling tower 1 proper was proposed. The cooling tower 1 was 'off loaded' by moving part of the load to the expanded cooling tower 2, allowing additional water flow to "key" exchangers served by the cooling tower 1.

Current Status

The Refinery has implemented several of the recommended projects and additional projects are schedule for installation in 1999 and 2000. The new internals and expansion of cooling tower 2, shifting of load from cooling tower 1 to 2 and the addition of the new booster pump has allowed 2 of the 3 constrained operating units to maintain capacity during the summer months. Planned and scheduled modifications will debottleneck the 3rd constrained unit.

The improvement on these constrained units has allowed increased gasoline and diesel production. The addition of surface area on process to cooling water exchangers and the recommissioning of the idle finfan significantly reduced the use of external water sprays and virtually eliminated flaring of light hydrocarbons due to cooling water system limitations. The remaining external water sprays should be eliminated next year when additional process to process and process to cooling water exchanger modifications are installed.


Paresh S. Parekh, P.E., PMP, is president of Unicade Inc., a consulting engineering firm in Bellevue, Washington. He has has 26 years of industrial, commercial, and international business management experience. He is an expert on chemical process engineering, systems engineering & classical project management techniques. His projects in industrial energy conservation and process optimization for pulp & paper, petroleum refineries, petrochemicals, heavy chemicals, chlor-alkali, plastics, food, electric utilities, and forest products industries have resulted in millions of dollars in operating cost reductions. He has presented at number of symposiums and technical meetings. Mr. Parekh is a past chairman of Puget Sound section of American Institute of Chemical Engineers and an active member of AIChE, PMI, AEE and ASHREA. Mr. Parekh has a Master of Science degree. in Chemical Engineering and is a registered professional engineer in the states of California and Washington. He is a certified Project Management Professional.

He can be reached at UNICADE INC., 13219 NE 20th Street, Suite 211, Bellevue, WA 98005. Tel. No. (425) 747-0353, Fax No. (425) 747-0316, e-mail: unicade@unicade.com, web site: http://www.unicade.com/.

John A. Latimer, P.E., is a Senior Project Engineer with Equilon Enterprises Puget Sound Refining Company, Anacortes, Wa. He has undertaken various Process and Project assignments in support of operations at the facility since joining Texaco in 1985. Latimer holds a BS (1984) in Chemical Engineering from the University of North Dakota. He is a registered engineer in the state of Washington. John can be reached at Puget Sound Refining Company, a Division of Equilon Enterprises, LLC, P.O.Box 622, 600 S. Texas Rd., Anacortes, WA 98221, phone: (360) 293-0863, e-mail: jalatimer@equilon.com.