An epiphany of energy savings for a petrochemical company by Visual MESA

Introduction:

Visual MESA, an Energy Real Time Optimizer (ERTO) gave an epiphany (a moment of sudden and great realization) in terms of ease of operation and energy savings for a petrochemical company in South East Asia in its intra and inter energy transactions. The overall plant complex is highly complicated consisting of two plants exchanging different pressure levels of steam and fuel, Boilers with varying capacity and fuel diet, steam import and export and power import. The outcomes after its implementation were new ways to operate the plant (seeing the invisible), better coordination and awareness among plant operators (feeling the intangible), and realizing the fruits of action by energy consumption reduction and economic savings (achieving the impossible). This article aims to present some of the evidences available in the form of actions taken from the data collected before and after taking the optimizer online. The understanding of this may prove to be of immense importance in energy optimization.

Seeing the invisible:

Plant operators always think that they operate the plant in the best possible way until their actions are reflected by a highly accurate thermodynamic model like Visual MESA which is running in parallel and real time. In one instance, there was an excess steam in the medium pressure header of Plant1. To keep the balance and avoid over pressuring the system, steam of 6.8 t/h was dumped to the atmosphere as shown in the below figure during normal operation.

MESA-Stream-Simulation.png

Venting steam is a wastage of energy and water. Although this strategy fixes the problem yet is not the optimal way to operate the steam balance. Visual MESA provided the optimal solution by cascading steam to a lower pressure header and then to the thermal deaerator. The below figure shows the delta view in the model, which is the difference between normal operation view and optimized operation view suggesting to decrease the vent indicated by -6.8 in t/h.

Visual-MESA-1.png

For the deaerator, the pressure limit was set between 1.5 Kg/cm2 and 2.5 Kg/cm2 and during normal operation the deaerator pressure was at 1.549 Kg/cm2 and during optimization the model suggested to push this to 2.5 Kg/cm2, thereby becoming a consumer for low pressure steam and also being within the pressure limits. The deaerator supplies the boiler feed water to the boilers and now since the boiler feed water is hotter, less fuel is needed in the boiler (reduction in fuel oil consumption at boilers of 0.8 t/h) to generate steam, and this constitutes the energy saving. The cost of 1 ton of fuel oil was 295 USD and the savings was 295x0.8 = 236 USD/h. Although considering the entire steam system, there can be other changes that offset this saving yet the optimized operation was better than current operation bringing in overall savings.

Visual-MESA-2.png

This was a new realization for the plant operators as otherwise the steam would have been wasted to atmosphere and the route which the model proposed and the subsequent fuel savings at the boiler was invisible to their eyes until then.

Feeling the intangible:

After taking the optimizer online it served as a watch dog for the company management. As MS Excel was used to generate reports and dashboard, it enabled users who are unfamiliar with model like the top management to take advantage of information which the software provided.

The operators and shift in charge between two plants began to talk with each other more often as coordinated strategies were required to reduce overall energy costs and at the end of each shift the model report was printed out to be handed over to the next shift in charge to make informed and agreed decisions.

Moreover the plant awareness increased tremendously by the way of knowing all the utility systems interactions, operating variables and associated constraints involved. Since the software is available offline as a standalone it had helped engineers with what if cases. Typical “What- If” studies done included evaluating turbine/motor switches, revamping equipment, and scheduling utilities for a plant wide shut down.

The software played a key role in tracking equipment performance by either estimating or calculating the non-measured consumptions like for example, boiler performance (i.e., efficiencies) can be tracked and maintenance scheduled when economically justified. Data quality identification by balloons which represent the header mass balance imbalances served to identify the bad actors (sensors or leaks).

Achieving the impossible:

The objective function of the optimization was to minimize the total energy cost of plant subject to the constraints. Until the commissioning of Visual MESA the realization of economic benefits due to optimal action was seldom possible as there was lack of tools available to compare the operation actions in real time and also it was always thought the plant was operated currently in the best possible way.

The economic benefits attributed to the use of Visual MESA real time optimization is by reducing the energy cost gap.

The Energy System Cost Gap (also known as “savings” or “delta cost”) corresponds to the economic benefit identified by Visual MESA and that can be captured through real time optimization (expressed in USD per hour). It is calculated as follows:

At each execution, Visual MESA calculates an “Energy System Cost Gap” as the difference between the “Simulated” or current energy system cost and the “Optimal” or best energy system cost determined in consideration of all defined constraints.

Energy System Cost Gap = Simulated Cost – Optimal Cost

Performance test was conducted in November and December of 2016 to establish the utility Optimizer base line (when no actions were taken by operators) followed by starting taking actions the Visual MESA recommended by Plant from 16th November onwards.

Baseline: From 4th November to 15th November 2016

Transition: 16th November to 1st December 2016

Reporting: From 2nd December to 10th December 2016


Economic-Benefits-1.png

A clear “step” can be identified between the period where no Visual MESA (VM) implementation was performed (Base line period) and the period where recommendations are implemented continuously.

Both baseline and reporting periods are close in time that makes the comparison more reliable.

Visual MESA (VM) calculated the cost gap based on current Fuel and Power prices. Cost Gap is calculated based on currently simulated and optimized utility cost of the Plant. All relevant cost calculation are histories in shistorian data base.

Baseline Period Average Energy Cost Gap = 946.7 USD/hr (Calcaulted by avareagin the cost Gap calcualted by VM)

Reporting Period Average Energy Cost Gap = 269.6 USD/hr (Calcaulted by avareagin the cost Gap calcualted by VM)

Energy Gap Improvement (USD/Hour) = Baseline Energy Cost Gap – Optimized Energy Cost Gap

= 946.7-269.6

= 677.1 USD/h

Plant Upstream factor = 8000 Hrs/year

Average Yearly Benefit = 677.1*8000=5,416,959 USD/Year

Therefore, the economic benefits captured for the real time optimization are:

Average Yearly Benefits = 5.4 million USD/year (rounded down)

The main source of economic benefits were reduction in propane injection (the costliest fuel) at the boilers and load balancing at boilers based on efficiency (maximum firing at the most efficient boilers to produce more steam) which are discussed in detail below. The other sources of benefits were reduction in letdown and maximization of extraction at compressor turbines, Demineralized (DM) water intake minimization for boiler feed water make up and swapping of pump turbines and motors.

1. Reduction in Propane Injection:

One of Main source of benefits is derived from the reduction of Propane injection in Fuel System. Propane is injected to make the shortage of Residual Gas also called off gas (RG) in the fuel System. However Visual MESA (VM) recommended the reduction of propane Injection and with increase in Fuel Oil (FO) make up to Fuel system. Propane was the costliest fuel followed by fuel oil and residual gas was free of cost.

Plot below is the extent of reduction in Propane Injection in Plant I and II Fuel System during performance test Period.

C3-Plant-I.png
C3-Plant-II.png

2. Load Balancing of Boilers based on Efficiency:

VM calculated the Boiler efficiencies real time based on ASME indirect method and then distributed the load among boilers to minimize the Fuel Cost.

Since PLANTI and PLANTII boilers are connected in same header and it is possible to exchange the steam between PLANTI and PLANTII utility system, VM calculated the boiler loads based on global Optimization considering the efficiencies of PLANTI and PLANTII boilers.

Plots below are boilers load during Base line and reporting period.

Boilers-Load.png

Plot below is efficiency of Boilers during base line and reporting period.

From the plots is it very clear that VM is recommending the continuous increase of load in more efficient boilers.

Boilers-Efficiency.png

From the analysis of the two plots above for Boiler load and Boiler efficiency, following is the conclusion:

  1. Boiler1C efficiency is continuously high and this Optimization is recommending to maximize the Boiler load.

  2. Boiler3A efficiency is higher so Optimizer is recommending maximization of Boiler3A load.

Closing thoughts:

Thus the objective to optimize the utility systems total operation cost of PLANT has been achieved by using Visual MESA Energy Real Time Optimizer model. An energy gap improvement of 677.1 USD/h was reported as the average energy cost gap between baseline line and implementation period. Most of the benefits has been achieved by using the most economic fuel to burn in the boilers and producing more steam from more efficient boilers. The Visual MESA energy real time optimizer gave operators actionable advice on how best to operate the complex with interactive utility systems to minimize utility cost on continual basis. In future there is also scope to model the utility system in great depth by including the emission limits for CO2 emissions and also closed loop optimization.