August 2020

Maintenance and Reliability

Improve stability of ammonia plant steam systems to reduce unscheduled plant shutdowns

Dyno Nobel runs a 2,300-metric tpd purifier plant in Waggaman, Louisiana.

Shukla, A., Dyno Nobel; Chandani, H., KBR; Thomerson, C., AVEVA

Dyno Nobel runs a 2,300-metric tpd purifier plant in Waggaman, Louisiana. It is one of the first onshore ammonia plants to be built in the U.S. in the past 20 yr, and one of the first built in Louisiana in a quarter century. The plant started operations in 2016. In its first year, the plant experienced frequent back-end trips that destabilized the steam system, resulting in a front-end trip that caused significant plant downtime.

In an ammonia plant, steam system stability is imperative for maximizing plant uptime. Due to the close integration between the process and the steam system (FIG. 1), any disturbance in the process can propagate to the steam system and affect the entire plant. The design intent is to keep the front end of the ammonia plant operating after a back-end trip. This is because a cascading trip will result in extended restart time, increased wear on the front-end equipment and loss of production.

FIG. 1. Block diagram of the ammonia plant.

Minimizing such cascading trips requires robust control of the steam system, which must quickly respond to a disturbance. A first-principle model, validated against plant data, can be used to evaluate the deficiencies in the existing control scheme. Exhaustive analyses of transient scenarios can then identify the modifications in control, logic/tuning parameters, valves, etc. Improved automated controls will result in minimizing required operator intervention immediately following an upset.

This article presents a case study for the Dyno Nobel ammonia plant, which experienced significant downtime due to steam system imbalance in its first year of operation. To understand the mechanisms causing the trips, and to determine potential remedial measures, a steam dynamic study (SDS) was performed for the plant. Specific examples are provided here to show the improvement of steam system control after implementation of the recommended modifications. Coordinated efforts of both the plant operations team and the simulation team, combined with systematic analysis and timely implementation, reduced unplanned plant shutdowns and increased plant uptime.

General SDS scope

A typical SDS to improve plant operations includes the following tasks:

  • Develop a dynamic model of the steam system (with major steam producers and users) and the control system
  • Evaluate interactions in the steam system
  • Determine the proper control tunings and load shedding strategies to prevent cascading trips.

Ammonia plant’s steam system

The ammonia plant’s steam system is shown in FIG. 2. It has three headers: high pressure (HP), medium pressure (MP) and low pressure (LP). The ammonia plant produces HP superheated steam, which drives the syngas and refrigeration compressor turbines. MP steam is extracted from the syngas and refrigeration compressor turbines to supply process steam, along with steam for the air compressor turbine, feed gas compressor turbine, miscellaneous pump turbines and heat exchangers. Excess MP steam is exported from the ammonia plant to users offsite. HP steam is generated by process heat recovery in three steam generators—one at the outlet of the secondary reformer and the other two at the outlet of the high-temperature shift (HTS) reactor and synthesis converter. HP steam is superheated to about 510°C (950°F) by additional process waste heat recovery in the HP steam superheater and reforming furnace convection section. This superheated steam is then supplied to the HP header for use. During the startup and emergency operations, MP steam is supplied from package boilers located offsite.

FIG. 2. The ammonia plant’s steam system.

First year of operations

The ammonia plant experienced several trips in its first year of operations, causing significant downtime. An analysis of the typical trip causes, performed by the operations team, showed that steam system imbalance was one of the major contributors (FIG. 3). In several instances, the back-end trips in the plant frequently destabilized the steam system, resulting in a front-end trip. These cascading trips resulted in extended restart time, increased wear on the front-end equipment and loss of production (FIG. 4).

FIG. 3. Analysis of typical trip causes.1
FIG. 4. Plant performance in the first year.1

To understand the mechanisms causing the trips, and to determine potential remedial measures, an SDS was performed on the plant. The following sections present the work method implemented for the study.

Work method

An operator training simulation (OTS) model for the plant was utilized for this study. The OTS system consisted of a rigorous process dynamic model of the complete plant, along with an emulated distributed control system (DCS) and safety instrumented system (SIS) controls. A five-step approach was adopted to study the trip events and develop potential solutions for preventing future occurrences. The steps involved included:

  • Conducting a root cause analysis (RCA) of the trip event
  • Producing a reproduction of the trip in the simulation model
  • Reviewing observations and recommendations with the plant operations team
  • Testing the recommended solutions in the simulation model
  • Validating recommendations with the operations team.

Performing an RCA

The Dyno Nobel plant operations team provided the author’s company with the historian data for key process variables, such as header pressures, steam flowrates, steam-to-carbon (S:C) ratio, as well as operator actions/event logs and first outs for several plant trips caused by steam system instability. An RCA was performed using the historian data to identify the potential causes of disturbance in the steam system when a trip occurred in the back end of the plant. Several hours of high-frequency historical data was trended and studied together with operator actions and DCS event logs. Three back-end trips were considered: the methanator, the syngas compressor and the refrigeration compressor. Of the three trips, a methanator trip is the most severe. It was found that maintaining adequate process steam flow, while managing the overall steam balance, was the key to preventing a complete plant shutdown in the event of a back-end trip.

Three factors influenced the process steam flowrate. These included the upstream pressure of the process steam control valve (i.e., the MP steam header pressure), the downstream pressure of the process steam control valve (i.e., the plant back pressure), and the control valves.

The original HP to MP letdown logic and front-end control system design in the plant did not provide enough time for operator intervention to prevent an upset of the S:C ratio. In most cases, this was the primary trigger for the cascading front-end trip (FIG. 5).

FIG. 5. S:C ratio during a back-end trip.

Reproducing the trip in the simulation model

The simulation model was baselined to an initial condition corresponding to plant steady-state operations. All the turbines were in service, and excess MP steam was being exported offsite. The transmitter ranges and tuning data for the key DCS controllers in the model were updated to match the as-built plant data. The shutdown logics in the model were also validated against and updated per the SIS in the plant. For model results validation, a scenario case for each plant trip considered was implemented in the model, based on the actions/event records provided by Dyno Nobel. The simulation results for the key parameters were matched against the plant trip data. For example, the following is a comparison of the simulation model and plant data for the methanator trip, which cascaded to a front-end trip.

FIGS. 6 and 7 show the comparison of plant and model results for the MP header pressure and process steam flowrate. The curve in blue represents plant data, and the curve in pink represents model data. The model results closely match the plant data.

FIG. 6. MP header pressure in the plant and simulation model.
FIG. 7. Process steam flowrate in the plant and simulation model.

Review observations and recommendations

Analysis of the plant historian data, along with reproductions of trip scenarios in the simulation model, led to some key observations. Based on these observations, the simulation team designed probable solutions that could be implemented at the plant without any significant capital cost impact. This was then reviewed with the plant operations team to get its agreement and feedback.

The letdown of HP steam to the MP steam header is essential for avoiding any cascading consequences from the initiating trip. Under normal operation, the MP steam export controller controls the MP header pressure by exporting excess MP steam offsite. In an upset condition, the MP header vent controller protects from overpressure on the MP header. The HP to MP letdown controller, along with a hand indicating controller (HIC) feedforward logic block, protects from under-pressure conditions on the MP header. The original HP to MP letdown logic implemented in the plant was not able to quickly stabilize the MP header pressure during a back-end trip. This always led to a rapid drop in the MP header pressure and to a subsequent upset of the S:C ratio. FIG. 8 shows the MP header pressure and S:C ratio observed in the methanator trip simulation. Reducing the dip in S:C ratio at around 70 sec was one of the goals of this study.

FIG. 8. MP header pressure and S:C ratio.

In addition to restabilizing the MP header pressure after the back-end trip, it is also critical for the operators to ensure adequate process steam flow. The process steam flowrate is influenced by both the upstream and downstream pressures of the process steam control valves. The MP steam header is the upstream side; the downstream side is the backpressure on the plant. During the back-end trip, while the MP header pressure dropped initially, the backpressure of the plant increased. As the available pressure drop across the control valve decreased, the flow controller increased the valve opening (FIG. 9). The response of the flow controller and vent (backpressure) controllers was not quick enough to help maintain the process steam flowrate.

FIG. 9. Pressure differential and valve opening.

Finally, to bring the steam system back in balance, quick operator actions are required to either reduce the amount of steam being used by the plant or to import steam from offsite. The process dynamics during the back-end upset and the original configuration of the control system did not provide operators with adequate time to respond.

Based on these observations, the following modifications were proposed to Dyno Nobel:

  • HP to MP steam letdown: The following were the recommended modifications for HP to MP letdown control logic settings:
    • Extend the minimum hold for the feed-forward signal. In the original logic, the feed-forward control would open to a preset value, hold for a certain time and then start to ramp closed. It was recommended that the minimum hold time should be extended to allow time for the MP header pressure to stabilize before the HIC starts ramping closed.
    • Increase the feed-forward letdown factor. In the original logic, the feed-forward control would open to a preset value based on a certain factor in the control logic. It was recommended that this factor should be slightly increased.
    • Include the current HIC value in the initial opening calculation. The HIC value should be increased by a value rather  than to a value.
  • Automatically decrease plant backpressure: It was recommended that, on a methanator or syngas compressor trip, to automatically reduce the setpoint on the vent upstream of the respective equipment to a preset value.
  • Trip close steam export: It was recommended that, on a syngas compressor trip, to close the MP steam export to preserve MP steam to the process. Steam import is always allowed through a check valve.
  • Automatically decrease the plant rate: It was recommended that, on a methanator or syngas compressor trip, to automatically set the plant rate to a reduced value. This would allow for an automatic reduction in process steam demand, coordinated with the gas and air flow.
  • Modify the process steam flow indicator controller split-range control: The process steam controller is a split-range controller sending output to valves A and B. The suggested modification was to start opening valve B earlier to compensate for the saturation of valve A at the top of the controller’s operating range.

In the event of a back-end trip, the main idea was to minimize the required initial operator intervention to provide the operators with enough time to take control of the situation and stabilize the plant.

Testing solutions in the simulation model

After reviewing the observations and probable solutions with the Dyno Nobel plant operations team, the recommendations were tested in a simulation model. The following case runs were performed after implementing the suggested changes:

  • Methanator trip, with gas venting upstream of the methanator
  • Methanator trip, with gas venting upstream of the low temperature shift (LTS) reactor
  • Synthesis gas compressor trip
  • Refrigeration compressor trip.

Case runs were performed to test each recommendation individually, and then in combination, to evaluate their effects. All of the suggested modifications improved the process performance, except for trip closing the MP steam export, which seemed to worsen the initial overshoot in MP header pressure. The overall effect of these modifications was to delay and reduce the magnitude of the MP header pressure dip and the resulting secondary dip in the S:C ratio. The automation enabled by the recommendations provided enough time for the operators to act and remedy the upset caused by the back-end trip.

Validating recommendations with the plant operations team

After testing recommendations for selected trip cases in the simulation model, the simulation team held a day-long review meeting with the Dyno Nobel plant operations team. The results of the simulation runs performed were presented to demonstrate the potential benefits of each recommendation. Detailed discussions were held to understand the impact of each modification, their feasibility, and the effort required to implement them in the plant.

Implementation in the plant

The Dyno Nobel team reviewed all suggested recommendations. Based on the feasibility of implementation and potential benefits, the team made three changes in the control systems logic:

  • Modified the HP to MP steam letdown logic
  • Automated lowering of setpoints on process vents
  • Automated plant rate reduction to 80%–85%.


The SDS for the Dyno Nobel ammonia plant was completed in August 2017. After implementation of the study’s recommendations, there was significant improvement in plant uptime in the facility’s second year of operations (FIGS. 10A, top; and 10B, bottom).

FIG. 10. Plant uptime in Year 1 (top) vs. plant uptime in Year 2 (bottom).1


The recommendations proposed from the SDS required low-cost improvements in the plant, which translated into significant benefits by improving plant uptime and also shortening the back-to-production time from approximately 40 hr for a complete plant trip to 20 hr for only a back-end trip. Additionally, reducing unnecessary heat cycles resulted in less stress on the front-end equipment, and preventing trips provided economic advantages.

Employing an SDS can help with a range of plant operational issues, control system validations and continual improvements of operating procedures. The work method described here also highlights the need for close engagement of the plant operations and simulation teams to achieve plant uptime improvement goals. HP


  1. “2018 DNLA PUG Presentation—Dyno Nobel Rev 2,” Dyno Nobel Presentation at Annual KBR Purifier Users Group Meeting, 2018.

The Authors

From the Archive



{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}