Pulp Mill Optimization No Longer Pulp Fiction

Home , Kappa number, Pulp mill, White liquor

Pulp mill optimization no longer pulp fiction

Producing high-quality pulp consistently with advanced process control

ABHIJIT BADWE, RAMESH SATINI – When Mondi – the global paper and cardboard producer – decided to modernize their largest pulp mill in Poland, they chose ABB as partners to help improve and optimize the mill’s two pulp production lines. For this undertaking, ABB devel- oped an advanced process control solution based on model predictive control and soft sensors. The modernization resulted in an improvement of 56 percent in pulp quality as well as standardized solutions that can now be offered to other pulp mills.

Pulp mill optimization no longer pulp fiction 53 1 Digester APC

Advanced controls (L3) 1) Pulp tracking 2) Soft sensors 3) Quality controls (MPC) a) Kappa control b) Residual alkali 4) Chip level (MPC) Chip bin Chip speed a) Level control OPT800 Cook/C b) Blow consistency 5) Reports & KPI

OPC

Supervisory controls (L2) 1) Alkali/wood and liquor/wood 2) Heating zones 3) Wash zone 4) HP feeder/production 5) H-factor control DCS*

Regulatory controls (L1) 1) Field devices interface 2) Process/safety interlocks Dual vessel operation

Blow flow * Part of DCS, not included in APC scope

Separating cellulose At the end of the process, the pulp is The raw material for cardboard and pa- squeezed out of the digester through an per is wood. The purpose of the pulping airlock called the blow line. The sudden process is to separate the valuable cel- depressurization results in the rapid ex- lulose fibers in wood from the lignin and pansion of the cellulose fibers, which n 2011, Mondi embarked on a project hemicellulose polymers that bind them. helps to separate them. They are then to look for ways to improve pulp qual- Traditionally, this was done by physically suspended in a liquid that is known to ity and chemical usage, enhance pro- beating the wood, but nowadays chemi- pulp workers as brown stock. I duction and reduce cost for its two cal methods are used – not least be- pulp production lines in the Mondi Swie- cause they better preserve the integrity The next stage of the process is to cie pulp mill. A number of factors played of the cellulose. “wash” the brown stock to remove the a role in this search. One was the com- cooking chemicals (now referred to as pany’s policy of reducing the environ- The process begins by stripping the black liquor) together with the degraded mental impact of its operations – and in wood of its bark, turning it into chips, lignin and hemicellulose. After progress- pulp mills this means cutting the use of screening the chips to achieve a roughly ing through a number of tanks, the ex- heat energy and chemicals. A second similar size, impregnating them with tracted liquid is reduced and its sodium factor was the requirement to improve chemicals and feeding them into a pres- and sulfur compounds recovered to the quality of the product. Mondi Swiecie sure cooker (the digester). make more white liquor. Meanwhile, the mostly makes containerboard, corrugat- clean cellulose pulp is bleached, if nec- ed packaging and industrial bags, which In the sulfate process used at Swiecie, essary, and then pressed and heated to are surprisingly high-tech products. the chips are mixed with a solution of remove water, after which it is ready to Thirdly, there were the universal goals of sodium hydrox- all industrial processes: increasing pro- ide and sodium duction volumes and decreasing produc- sulfide (known as One of the first things that tion costs. white liquor), pres- surized and heat- the newly formed Mondi In 2011, Mondi Swiecie awarded ABB a ed to a temper contract to improve the quality and effi- ature of about Swiecie did was look for ways ciency of its pulp mill. To better under- 160 °C with steam to improve the pulp mill’s stand the challenge involved in this task, from coal-fired boil it pays to take a closer look at how the ers. After a few production process. pulp process works. hours, the material takes on the consistency and color of porridge and be cut and rolled or bailed for further Title picture the amount of alkali in the sodium processing. Advanced process control at the largest pulp mill in hydroxide decreases as it is turned into Poland has increased productivity – and stabilized sulfates and carbonates as a result of the processes in the cooking and pulp washing lines in a way that conventional methods of DCS reacting with the lignin in the wood. control could not.

54 ABB review 1|16 2 Control overview – washing APC In 2011, Mondi

Dilution factor control

Advanced process controls Swiecie awarded Filtrate tank level control 1) Pulp tracking Digester washing control 2) Grade change

3) Soft sensors ABB a contract to 4) Tank level control (MPC) DD washer1 5) Conductivity controls improve the quality Digester DD washer2 (MPC) at each stage OPT800 Wash 6) Process KPI Diffuser 7) Defoamer controls and efficiency of Pressure Pressure its pulp mill. diffuser 1 diffuser 2

Supervisory control 1) Production rate control 2) Dilution factor

3) DD washer/diffuser DCS controls

have been cooked. This number is diffi- The challenge was cult to control for two reasons. Firstly, because all the chips that are fed in have to make the pro- different moisture content and physical cess work more characteristics. Secondly, the Kappa number cannot be physically measured efficiently and to in the digester, but only after the chips have passed through it and entered the higher tolerances blow line. This is a problem because it is essential to know what the Kappa num- without replacing ber is before this point. Like anything anything except else that goes into an oven, the pulp has to stay there for just the right amount of the control system time – long enough to yield as much cellulose as possible, but not so long as to that operated the break down its physical structure. existing equipment. Therefore, to maintain a steady process with minimal variations in the quality of pulp, the Kappa number has to be ar- How to improve the process rived at by taking continuous measure- The challenge was to make the process ments of the various process variables work more efficiently and to higher toler- before the chips enter the digester and ances without replacing anything except feeding these numbers into a mathemat- the control system that operated the ex- ical model (or soft sensor) that considers isting equipment. The key to this was to the multiple nonlinear process effects. optimize the operation of the two lines’ digester and brown stock washing stag- This is what ABB’s solution does ➔ 1. es. The first step was to audit and ana- The advanced process control scheme lyze the control loops, then develop an employs a soft sensor – based on ABB’s advanced process control (APC) action Inferential Modeling Platform – that plan based on model predictive control yields soft measurements of the Kappa (MPC) and soft sensors. number from a series of process variable measurements. These measure- The cooking process ments are tracked using a tracking For the digester, the key variable for function that creates a virtual model of product quality is the Kappa number – the chips on their journey through the this measures the lignin level in the pulp digester. Along the way, measurements and thereby indicates how well the chips are derived from the chip’s characteris-

Pulp mill optimization no longer pulp fiction 55 ABB’s APC solu- 3 APC implementation approach tion maintains the Diagnose Implement Sustain chip level by opti- Service Service mally manipulating Pulp Performance MPC the digester bot- tracking monitoring - Blow - Kappa MPC - Control loop - Chip bin - Quarterly consistency - Dig level MPC - Performance audit - Impregnation audit tom scraper speed - Kappa model - Step test test run - Analysis, vessel - Benchmarking - Install IMP - Develop - Case study findings - Digester results - Data models - Success story and the pulp flow - Action plan vessel - ServicePort collection - Deploy MPC out of the digester. Process Soft sensor Results fingerprint models report

4 Reduction in Kappa Variation 5 Reduction in residual alkali variation (2 Sigma) (2 Sigma)

8 1.50

7 1.25 6 1.00 5

4 0.75

3 0.50 Line 2 kappa Sigma 2 Residual alkali 2 Sigma 0.25 1

0 0 Before APC After APC Before APC After APC

tics and these are fed into the model to Another important variable in the contin- predict the Kappa number in every zone uous cooking process is the level of of the digester. The soft sensor is de- chips inside the digester. Variations in ployed online and yields real-time virtual this chip level leads to nonuniform cook- measurements of quality variables (such ing, disturbances in the overall liquor bal- as the Kappa number) that are then used by a The first step was to audit model predictive controller to opti- and analyze the control loops, mize the cooking then develop an advanced process, to de- cide the length of process control action plan time the chips are exposed to a based on model predictive given concentra- control and soft sensors. tion of effective alkali and to keep the concentration of alkali the same ance of the process and nonuniform pulp throughout the digester vessel. The flow at the digester outlet. ABB’s APC APC is also able to maintain a consis- solution maintains the chip level by opti- tent production rate of pulp. Finally, a mally manipulating the digester bottom log of the data that has been collected scraper speed and the pulp flow (or blow is packaged into concise reports, com- flow) out of the digester. plete with measurements of key performance indicators.

56 ABB review 1|16 6 APC stabilizes chip level The approach followed to achieve Mondi’s objectives was based on ABB’s Advanced Services philosophy – diagnose, 6a Chip level with no APC 6b Chip level with APC implement and 30

25 sustain.

10 Dig Chip Level

0 Before APC After APC

6c Process stability is greater when APC is on. 6d APC leads to reduced digester chip levels.

While the APC is handling these complex the digester, the data is gathered on a measurements, making real-time predic- continuous basis so that soft sensors tions about the process and implement- can give predictive readings of what is ing optimal control actions, the ABB taking place. Extended Automation System 800xA distributed control system (DCS) handles The aim in this stage of the process is to the basic controls such as liquor and control the levels in the numerous fil- chip flows, temperature, pressure, etc. It trate tanks so that just the right amounts also performs the vital job of controlling of fresh water are added for each ton of the H-factor – that is, the rate at which brown stock to be washed (this is known the lignin is being dissolved. As this is as the dilution factor). If an optimal val- largely a function of temperature, the ue for this variable can be continuously amount of heat applied to the digester solved for, then the maximum quantity has to be closely controlled. A variance of the cooking chemicals can be recov- from the optimal of just a couple of ered and the minimum amount of ener- degrees can make a big difference to gy used in the evaporators. The alkali the quality of the pulp. losses will also be reduced. This is achieved by the brown stock washing The obvious advantage of the APC and APC ➔ 2, which computes the optimal the DCS is that the outcome meets all of dilution factor at each washing stage the customer’s requirements for this step based on pulp conductivity measure- of the pulping process. The fact that just ments at various locations in the pro- the right amount of steam has been add- cess; at the same time, it ensures that ed and just the right volume of white the levels in the filtrate tanks are within liquor or alkali injected means that as their specified ranges. As with the much waste as possible has been digester, the information gathered squeezed out of the process. And as the throughout the operation is presented maximum amount of highest quality cel- to the control room in the form of key lulose has been produced, the quality and performance indicators. quantity criteria have been met as well. The implementation methodology The brown stock washing process The approach followed to achieve Mon- The second key area where optimization di’s objectives was based on ABB’s Ad- has been achieved through an APC sys- vanced Services philosophy – diagnose, tem is the brown stock washing. As with implement and sustain ➔ 3.

Pulp mill optimization no longer pulp fiction 57 While the APC is 7 APC stabilizes blow flow handling these complex measurements, making real-time predic- tions about the process and implementing optimal control actions, the ABB System 800xA DCS handles the basic controls. The diagnose phase involved performing dent in using the APC and were able to a detailed audit of the process and con- adapt their working and training practic- trol loops at Mondi Swiecie, which led to es to make best use of it. the development of a specific action plan for implementing the APC system. Once the entire system was up and run- ning, it was possible to arrive at some The implementation phase involved the quantitative measures as to how well the deployment of the pulp tracking func- objective had been fulfilled: tion in the DCS followed by the develop- − Variations in the Kappa number in the ment of online soft sensors to model digester were reduced by 56 percent, the Kappa number. Subsequently, step meaning that the APC has helped tests were carried out on the process to produce pulp of a consistently high develop mathematical models for use in quality ➔ 4. the model predictive controller, which − Variations in residual alkali fell by 48 was then commissioned, tuned and percent, implying a better utilization of handed over to the mill operators. white liquor in the cooking process ➔ 5. − Variations in the digester chip level The APCs are currently in the sustain were cut by 40 percent ➔ 6, resulting phase, in which ABB helps Mondi main- in reduced variations in the blow tain the optimal performance of the flow ➔ 7. This has led to much more APCs by performing quarterly audits, stability in the pulping and washing remote monitoring and online retuning. processes.

A similar approach was fol- The obvious advantage of lowed for the implementation of the APC and the DCS is that the APC in the washing area. the outcome meets all of the customer’s requirements The result This work was for this step of the pulping. carried out in phases between August 2011 and December 2013, by − A 30 percent reduction in alkali losses which time the system was completely in the washing proces was installed and tested. During this process, achieved ➔ 8. the pulp mill’s engineers were kept fully engaged to make sure they were confi-

58 ABB review 1|16 8 APC reduces alkali losses in brown stock washing

Upper limit = 12 kg/ADT 125

100

75 Before APC: 48% of total data above upper limit 50 Occurrence: raw data Occurrence:

0 5.0 7.5 10.0 12.5 15.0 17.5 20.0

Alkali loss (kg/ADT)

8a Before APC

60 Target shift: from 11.5 to 8.9 kg/ADT: ~3kgs per ton of pulp washed 50

40 After APC: 30 8% of total data above upper limit 20 Occurrence: raw data Occurrence:

0 5.0 7.5 10.0 12.5 15.0 17.5 20.0

Alkali loss (kg/ADT)

Histogram data Normal curve ADT = air dry ton

8b After APC

− The volume of fresh water used in the siderations was to have a proven APC washing area and consequently solution on top of the DCS – the Mondi steam consumption by the evapora- Swiecie project provided this. tors were both reduced. − Washing filtrate and blow tank levels were stabilized, resulting in a reduced carbon dioxide and effluent load.

The whole process of solving the optimization problem and having that solution validated by the mill itself has made it possible for ABB to offer standardized solutions to other pulp mills – these solutions are now called OPT800 Cook/C for continuous digesters and OPT800 Wash for brown stock washing.

Finally, the APC solution has proved to Abhijit Badwe be a door opener not only for other APC Ramesh Satini opportunities but also for System 800xA ABB Pulp and Paper Control Systems DCSs as well. One recent example is a Technology Center DCS order for the world’s largest pulp Singapore and paper mill, which is being set up in [email protected] Indonesia. One of the key customer con- [email protected]

Pulp mill optimization no longer pulp fiction 59