You are on page 1of 4

PLANT/PROCESS OPTIMIZATION SPECIALREPORT

Rethink planning and scheduling’s


role in refinery optimization
New tools improve prediction of unit feed qualities and yields
M. SWENSEN, Valero Energy Corporation, San Antonio, Texas;
and C. ACUFF, M3 Technology, Houston, Texas

P
lanning and scheduling are the core business functions the refinery, operating rates, parameters and yields on each pro-
whereby refiners adjust the entire manufacturing process to cess unit for each day. It also includes timing of batches, recipes
improve profits and achieve better results. Whether it’s used at required to produce different product grades and the detailed
a small refinery or a large one, planning and scheduling can have a product shipment schedule into the future.
major impact on profitability. A case in point is Valero Energy Cor- Scheduling also makes use of offline built-in optimizers for
poration’s 14 refineries with crude and feedstock throughput ranging maximizing profit for crude and product blending. The blend
from 90,000 barrels per day (bpd) to over 300,000 bpd. Improved optimizer can be used to optimize a single crude or gasoline blend,
planning and scheduling can have a positive impact within the refin- or it can be used to develop optimum recipes for a sequence of
ery as well as across the company’s supply chain. With today’s crude blends through time (multi-period optimization). In the case of
prices pushing past $100/bbl, a small “per barrel” impact can be very products blending, the optimizer maximizes profitability as well as
significant. Regardless of size and economies of scale, the specifics minimizes quality giveaway, all while meeting target specifications
of planning and scheduling are applicable throughout the industry. and managing inventory balances of component tanks.
This article will cover the challenges, solutions, results and benefits In short, scheduling/blending is the development of the detailed,
of planning and scheduling in refinery optimization. executable plan to implement the operating strategy developed in
the planning process, running the refinery up against identified
Planning and scheduling/blending defined. The pro- physical or economic constraints to maximize profitability.
duction planning system provides the strategic direction regarding
which crudes to purchase and in what proportions, the need for Solution for Valero’s diverse refining operations.
additional feedstocks, the operating rates and conditions of each Beginning with just one refinery in 1997, Valero has grown
process unit, and the optimal product slate. A linear program (LP) through acquisitions and mergers to become North America’s
is the primary tool used to vary all of these parameters to optimize largest independent refiner, with 14 refineries having a combined
refinery profitability over the planning period, taking into account production capacity of more than 2.6 million bpd. Valero’s refin-
projected market prices, unit operating constraints and avail- ing network stretches from the US West Coast, Mid-Continent
ability, and expected demand for products in various channels of and Gulf Coast to Canada and the Caribbean. Such a diverse set
trade. The planning tool assumes a continuous, steady supply of of assets produces many unique challenges. For example, crude
crude and feedstocks, and a continuous demand for products and supply can be ratable, locally produced crude or foreign imports.
stable refinery operations over the period. It does not model the Refineries might be landlocked or have direct access to waterborne
build or draw of inventories. Based on these assumptions, the LP movements. Storage tanks can include a mixture of offsite/third-
produces the optimal “period-average” operating plan. party terminals and onsite refinery tanks. Tank stratification or
In reality, refineries do not operate on averages. Crude and partial mixing and the use of running-gauge tanks can have a big
feedstocks arrive in batches, and product is shipped in batches. impact on operations. Tracking pipeline line-fill volumes and
While the refinery runs continuously, individual units may run compositions can be essential in some situations and unnecessary
differently for periods of time for various reasons. Planning is not in others. Product blending facilities range from batch blend-
event based, which is why scheduling is needed. Scheduling takes ing and run-down blending to configurations with individual
all of these discrete events into account with the goal of imple- component tanks, online analyzers, and online batch quality
menting the operating strategy developed in the planning process. control. Product specifications include California Air Resources
Scheduling provides dynamic responses to changes as they Board (CARB) grades, unique one-time export grades, Refor-
occur (called event-based capabilities). Scheduling details out the mulated Blendstock for Oxygenate Blending (RBOB) and other
events as they will occur in the refinery, including such things as grades such as Conventional Blendstock for Oxygenate Blending
the timing of receipts and the tanks into which crude and feed- (CBOB) and conventional grades.
stocks will be received. The scheduling tool predicts the resulting In mid-2008, Valero initiated a project to improve and stan-
tank composition and qualities. The schedule reflects details dardize refinery scheduling and blending tools, as well as work
such as how various intermediate streams will be routed within processes. After completing software selection, end-user align-
HYDROCARBON PROCESSING JUNE 2011
I 55
SPECIALREPORT PLANT/PROCESS OPTIMIZATION

ment, and implementation plan development, the first refinery database-backed scheduling system. As of early April 2011, imple-
implementation kicked off in early March 2009. The main focus mentation is nearing completion across all Valero’s US refineries.
of the project was the deployment of refinery scheduling and a
multi-blend optimizer software as a new standard crude, feed- Refinery optimization is a collaborative effort. At
stock and product scheduling/blending platform. The new tool Valero, many groups and individuals participate in the opti-
would replace multiple non-standard, non-integrated spread- mization of a refinery, including the refinery planner, refinery
sheets at each refinery. The scheduling/blending platform captures schedulers (e.g., crude, process unit, distillate/other products,
the uniqueness of each refinery and its associated logistics while and gasoline blenders/schedulers), crude and feedstock planning,
allowing for central support of models and reports, development product planning, transportation, supply and trading, refinery
and sharing of best practices, and data mining/reporting from a operations, and refinery oil movements.
In the past, scheduling systems were restricted to a scope of
“inside the fence” of the refinery. Today, scheduling systems
Alkylate span across the supply chain, terminals, refineries and organiza-
1st blend Product
header tanks tional groups. This enables a collaborative effort, better visibility
and additional points of adjustment for improving economics.
Component
tanks Adjustments are seen by multiple groups, enabling faster response
Product and awareness.
tanks
Reformate Scheduling integrates data, technology and people. The more
2nd blend
header feedback and comparison that is enabled, the more effective the
optimization becomes. For example, the planning model’s pro-
Product
Component tanks
cess unit sub-models are shared with the scheduling system. The
tanks scheduling system compares simulated unit performance vs. actual
3rd blend unit performance. The refinery scheduler can apply biases to
Naphtha header
Product compensate for sub-model inaccuracy. The planning group then
tanks receives feedback on sub-model accuracy, and this information
Process Manifold Manifold is then used to tune submodels, ultimately benefiting both the
units
planning and scheduling tools.
FIG. 1 Rundown blending optimization with three blend headers. Timely updates of shipments and receipts (i.e. nominations)
from the supply and trading group are reflected in the refinery
scheduling system for the crude, feedstocks and product’s schedul-
ers. Supply and trading receives feedback on actual arrival/depar-
ture times, demurrage events, quantities and material properties.
Refinery operations and oil movements receive daily operating
orders from the refinery schedulers, including direct import of
receipts, shipments, blends and tank transfers into oil movement
tracking systems. In short, scheduling is the hub that coordinates the
Drying Technology contributions of many to optimize the refinery.

Rundown blending optimization. Fig. 1 shows the sche-


matic for rundown blending optimization. Rundown blending
involves multiple process unit streams as continuous feed com-
ponents into blend headers. The blend headers may also pull feed
from typical component tankage. As depicted, at any point in
time, three blend headers receive feed from any of the component

Primary Pipeline A Pipeline B Pipeline C Finished


crude Refinery product
blending tank
terminal farm
Marine Pipeline D

CARBO
N Secondary
liftings

DR O
You
ficar HY
t io n crude
terminal

eci
Pipeline E
Ke
ep Distribution
sp terminal

Wa
te r
CONTR
O L .. .
u nd
er Simulation scope includes terminal activities,
pipeline batches, refinery, and distribution

FIG. 2 Refinery supply-chain logistics.


www.newton-s.com
Select 165 at www.HydrocarbonProcessing.com/RS
56
PLANT/PROCESS OPTIMIZATION SPECIALREPORT

tanks and process unit rundown streams. Rundown streams must implementation of a more robust tool as well as the improvement
be completely consumed in the three blends, since there is no other of associated work processes. Several observed benefits were noted:
possible destination for these streams. Volume of each blend can be • Data from the process historian, lab information system,
fixed or allowed to vary. The blend optimizer solves the group of movement systems and commercial nomination schedule system
three blends together to meet specifications for each grade, mini- are integrated using web services and the supplier’s integration tool
mize quality giveaway and maximize profits. The optimizer can • Tracking of crude properties in each tank and in the crude
also optimize sequential groups of blends in a multi-period optimi- feed blends provide prediction of unit feed qualities and yields:
zation. While Fig. 1 shows three blend headers, the solution is user o At one Valero refinery, crude total acid number (TAN)
configured for any number necessary for the site. While depicted was increased an average of 0.2 points by better predicting the
with gasoline blending, it can also be configured to optimize diesel TAN of a particular side draw.
or fuel-oil blends that are often managed as rundown blends. o At another refinery, use of the blending tool for crude

Refinery supply-chain logistics. Managing supply and Crude and product blend comparison
demand is a requirement for achieving overall refinery optimiza-
Deviation analysis
tion. Without doing so, a refinery can be constrained on blend Actual blend results
feedstocks, production rates, or shipment of finished product. At …
Valero, the refinery supply-chain network includes a variety of Scheduled blends
Blend
performance
docks, terminals, tank farms, refineries and pipelines. Fig. 2 shows Destination facility multi-dimensional
the supply-chain scope for a marine-based refinery with multiple Component facilities comparison
Recipe
pipelines and terminal assets. These assets along with the produc- Heel(s)
tion capability of the refinery are visually shown in a model mak- Volumes, properties
ing it easy to see the possible inventory limitations, throughputs
Blend #1 Blend #2 Blend #3 Blend #4 Blend #5 Blend #6
and constraints across the refinery supply chain network.
Pipelines A, B, C, D and E reflect real logistical challenges to
maintaining balance across the network. Each pipeline has unique Single blend Multi-blend Long-term blend
optimizer optimizer optimization using
linefill, minimum/maximum rate constraints, multiple inlets/out- aggregate tankage
lets, and default throughput rates. Crude pipeline chains, such as A
and B, can work sequentially with or without the secondary crude FIG. 3 Single- and multi-blend performance analysis.
terminal. Likewise, the product pipeline chain has the flexibility to
provide product to a secondary distribution terminal or to a com-
mon carrier pipeline (E). The solution is to simulate the pipeline
batches, terminal activities and refinery production/consumption
across the supply-chain network to see the big picture of where
each asset stands. Volume and properties for each pipeline batch
and pipeline segment are available at any point in the simulation.

Blend performance analysis. To use either the single blend


optimizer or the multi-period blend optimizer, the blending/
scheduling tool must accurately predict the properties and quali-
ties of product batches. Constant vigilance is required to ensure
that blend property prediction correlations are properly tuned.
In Fig. 3, scheduled blends that have been optimized using the
blending tool’s single- and multi-blend solvers are transferred to
refinery operations for execution. Once the blends are executed,
results are brought back into the blending tool for comparison
and analysis. The goal is to measure the deviation (i.e., between
predicted and actual product qualities) in order to tune the blend-
ing correlations.
Inspect what you expect! Blend performance analysis provides
an easy way to compare scheduled vs. actual results. For example,
the actual recipe can be applied to the scheduled component qual-
ities. Or the effects of the scheduled recipe can be analyzed with
the actual component qualities (instead of simulated qualities).
Comparisons include simulated vs. actual for: (1) blend recipe, (2)
component quality, (3) destination tank quality including heel,
and (4) batch quality, excluding heel. Deviations are automatically
calculated across the multiple dimensions. This makes tracking
down the root cause for the deviation much faster.

Benefits across the organization. Benefits from the sched-


uling/blending project have materialized, both as a result of the
Select 166 at www.HydrocarbonProcessing.com/RS
57
SPECIALREPORT PLANT/PROCESS OPTIMIZATION

blend optimization allowed better utilization of all process units o One scheduler at a marine-based refinery noted that they
during the higher-demand summer gasoline season used to be able to tell oil movements whether a ship could be
o At yet another refinery, the blending tool’s virtual blend brought to the dock tomorrow, but can now tell them what time
feature allows optimized crude recipes to include crude cargoes tomorrow to bring it in.
that are yet to be received into tankage. Valero expects to capture more benefits going forward with
• Cargoes of lower-quality intermediate feeds no longer result the completed scheduling/blending platform implementation.
in surprises for unit operations because receipts can be simulated Valero will leverage best practices with the improved scheduling/
before being used. blending platform and corporate commercial scheduling tools.
o Rate on one resid cracking unit would typically fall by The integrated refinery scheduling/blending tool has become the
6,000–8,000 bpd when impacted by lower quality feeds. Now, foundation for better communication, and it serves as the conduit
due to better scheduling and blending of feeds, the production for all information related to planned plant operations as well as
rate impact is typically only 2,000–3,000 bpd. for crude, feedstock and product movements. HP
• At a West Coast refinery, schedulers used to work with three
separate tools: a legacy DOS-based blend optimizer, a spreadsheet Mark Swensen is director of supply-chain optimization for
scheduling tool, and the CARB predictive spreadsheet model, Valero Energy Corp. He has been with Valero for nine years. Mr.
The blend otimizer can use “passing CARB” as a product quality Swensen’s other roles at Valero include strategic planning, mergers
specification. Now all three tools are combined into one. and acquisitions, Latin American business development, and alterna-
tive energy business development. Before Valero, his employment
• Common gasoline blend property predictions are available included business strategy consulting and roles as process engineer and startup/com-
by integrating Valero’s proprietary blend equation library into missioning engineer. Mr. Swensen earned a BS degree in chemical engineering from
both the planning tool and the scheduling/blending platform. Brigham Young University and an MBA degree from the Darden School of Business at
o Combined with the blending tool, it optimizes profit- the University of Virginia.
ability while still managing inventories was accomplished.
o During Reid vapor pressure transition, one Gulf Coast Craig Acuff is the business development director with M3 Tech-
refinery had always required purchases of a high-cost blend stock. nology’s SIMTO (scheduling and simulation) and M-Blend (single/
They were able to manage the most recent transition with no pur- multi-period blend optimization). Mr. Acuff has 24 years of experi-
chases needed. ence and has been involved with implementing refinery systems for
achieving business process improvements, along with assisting many
• Tracking and managing dock availability and utilization end users with making the transition to best practices. Mr. Acuff earned a BS degree
allows better management of inventories and minimization of in mathematics from Oklahoma State University and an MS degree specializing in
inventories. computer science from the University of Central Oklahoma.

LIVE WEBCAST
presents . . .
Thursday, June 23, 2011
11 a.m. ET / 10 a.m. CT

“Don’t Pay At the Pump: Four Smart Steps Refineries Can


Take to Lower Production Costs”
While consumers are used to buying gasoline at the pump, many re½neries are “paying at the pump” in a negative way that
drives up production costs. Optimizing the ef½ciency of API pumps—which account for 60 percent of the motor energy used
in typical re½ning operations—is one of the simplest ways to boost reliability and save energy on production, so that you can
sell more to customers.
In this webcast, Mike Pemberton, an energy ef½ciency expert for ITT Goulds Pumps, will give practical advice supported by
speci½c examples from re½ning operations for Chevron, Shell and other major companies. Attendees will learn:
• How to select the right-sized pump—and why many are oversized, with some undersized
• Tips for identifying the 10 percent of pumps that offer the fastest payback from optimization
• Facts and fallacies about smart controls
• How asset management is helping smart re½neries to boost bottom-line performance
This webcast will deliver “news you can use” for anyone who oversees pumps, energy utilization and reliability—including
re½nery managers, process engineers, production supervisors, reliability engineers and purchasing.

ITT Speaker: Mike Pemberton

REGISTER at
www.HydrocarbonProcessing.com
Select 167 at www.HydrocarbonProcessing.com/RS

You might also like