You are on page 1of 6

Mixed Model Assembly Sequencing at Toyota: Project Update

by

Bo Liu,

Yudhisthir Paudel

Instructor: Leonardo Bedoya, Ph.D

EN 504 Scheduling and Sequencing

COLORADO STATE UNIVERSITY-PUEBLO


BACKGROUND:

Toyota is a dominant and one of the most successful auto-manufacturers in the world at modern times.
This success came to Toyota because of its very efficient scheduling method which aims to reduce
unnecessary functions within the facility and the production and assembly process. There was a natural
pressure for the company to stay competitive with the western auto makers and hence it came up with
Just-in-time production system, also well known as Mixed-model assembly sequencing. Because the
production system was created from actual practices within Toyota, it emphases strongly on practical
effects and implementation over theoretical analysis. 1

The basis of the Toyota production system is the absolute elimination of waste. The fundamental
objective based on the system are-

 Just in time
 Autonomation, or automation with a human touch. 2

In the mixed-model assembly line, different models are introduced in different sequences because of
different goals or purpose for controlling the line. The two different goals are-

1. Leveling the load (total assembly line) on each process within the line (Goal 1).
2. Keeping a constant speed in consuming each part on the line(Goal 2). 3

The first goal is also considered in Toyota’s sequencing program, it is incorporated in the solution
algorithm which mainly considers the second goal. The two common algorithms used by Toyota to
achieve these goals are Goal Chasing Algorithm-I and Goal Chasing Algorithm-II.

Goal 1 considers possibly different operation times of various models at each assembly station and
attempts to avoid problems such as delay or line stoppage caused by successively longer operation
times. Okamura and Yamashina (1979) developed a sequencing method for mixed-model assembly lines
to minimize line stoppage. Yano and Rachamadugu (1991) addressed the problem of sequencing mixed-
model assembly lines to minimize work overload.

Goal 2 (to keep constant rates of part usage by the line) is considered in the Toyota production system
to be more important than Goal 1 (Mondon, 1983), and it is incorporated in the solution algorithm
which mainly considers the solution algorithm-I. This is because smooth production would not be
realized without achieving this goal. To keep constant rates of part usage implies an objective function
that minimizes the total variation between the actual production rate and the ideal (constant)
production rate for every model produced on the assembly line. 3
INTRODUCTION:
In the assembly system used in Toyota, great importance is given to minimize the variation in production
quantities or conveyance times at preceding processes and also the respective work-in-process
inventories. For this purpose, the consumption speed needs to be kept as constant as possible. The
notations used in the sequencing method are defined below.

Q = Total production quantity of all products Ai (i= 1,…..,α)

α
¿ ∑ Qi , (Qi production quantity of each product Ai)
i=1

Nj = Total necessary quantity of the part ai to be consumed for producing all products Ai (i=1,….α; j=1,
….,β)

Xjk = Total necessary quantity of the part aj to be utilized for producing the products of determined
sequence first to Kth.

Using the above notations, the following two values can be developed:
Nj/Q = Average necessary quantity of the part aj per unit of a product.

K . Nj
= Average necessary quantity of the part aj for producing K units of products.
Q

Let’s define two points Gk and Pk as follows.

Gk = (K.N1/Q, K.N2/Q,……K.Nβ/Q),

Pk= (X1k, X2k,…..,Xβk)

The point Gk and Pk must be as close as possible to assure the constant speed of consuming each part. If
we say Dk is the distance between the two points, we will minimize the following function:

β
Dk = ‖Gk−Pk‖=
√ ∑(
j=1
K . Nj
Q
−Xjk )
2

This is the basic idea behind the goal-chasing method for the mixed model assembly used by Toyota. 3

The algorithm developed and used based on this idea is presented below.
Goal Chasing Algorithm-I

Let,

bij= Necessary quantity of the part aj (j=1,….,β) for producing one unit of the product A i (i=1,…,α).

Then,

Step 1: Set K=1, Xj,k-1 = 0 , (j=1,…, β), Sk-1= {1,2,…, α}.

Step 2: Set as Kth order in the sequence schedule the product A i which minimizes the distance Dk. The
minimum distance will be found by the following formula.

Dki* = min
i
{ Dki } i Є Sk-1

β
where Dki =
√ ∑(
j=1
K . Nj
Q
−Xjk −bij)
2

Step 3: If all units of a product Ai* were ordered and included in the sequence schedule, then

Set Sk = Sk-1 – {i*}.

If some units of a product Ai* are still remaining as being not ordered, then set S k = Sk-1

Step 4: If Sk = ф (empty set), the algorithm will end.

If Sk ≠ 0, then compute Xjk = Xj,k-1+ bi*j (j=1,…β) and go back to Step 2 by setting K = K+1.

Example Problem

An example problem has been solved using this algorithm in matlab.

Suppose the production quantities Qi (i=1,2,3) of each product A1, A2 and A3, and the required unit b ij
(i=1,2,3;j=1,2,3,4) of each part a1,a2,a3 and a4.

Products Ai A1 A2 A3
Planned Production 2 3 5s
QuantityQi
Product Parts aj a1 a2 a3 a4
s Ai
A1 1 0 1 1
A2 1 1 0 1
A3 0 1 1 0

The matlab code used for the solution is in Index-1, which gives us the sequence schedule.

We will either develop an alternative algorithm for the goal chasing method or apply this algorithm to a
different manufacturing environment other than automobile manufacture, or do both options.

References:

1. Taiichi Ohno (former vice president, Toyota motor corporation); Toyota Production System,
Yasuhiro Monden, 1983
2. Toyota Production System, Taiichi Ohno, Productivity Press, USA, 1926,p.4
3. Toyota Production System, Yasuhiro Monden, 1983, p.182

Index:1

Matlab code for solving the given example

%% Goal-Chasing Method-1
PPQ=[2 3 5];
PC1=[1 0 1 1];
PC2=[1 1 0 1];
PC3=[0 1 1 0];
bj=[PC1;PC2;PC3];
TNQ=PPQ*bj;
TPQ=sum(PPQ);
NjQ=TNQ/TPQ;
K=1;
Xjk=0;
Seq=[];
test=[];
flag=1;
while flag
D=[];
for i=1:3
a=sqrt(sum((K*TNQ./TPQ-Xjk-bj(i,:)).^2));
D=[D a];
end
[n,m]=min(D);
test=[test,m];
c1=sum(+(test==1));
c2=sum(+(test==2));
c3=sum(+(test==3));
b=[' A' num2str(m)];
Seq=[Seq,b]
Xjk=Xjk+bj(m,:)
K=K+1;
if c1==PPQ(1)&&c2==PPQ(2)&&c3==PPQ(3)
flag=0;
break
end
end

You might also like