How to find a stable response with state space and PID controller?

I need to find response stable y=[108;66.65;0] with State space module and multivariable 'PI' controller, but with my representation the response is unstable. It is correct what I have done it? And, how to find a stable response with state space and multivariable 'PI' controller?

State Space Parameters: A=[-2.509e-3 0 0;6.940e-2 -0.1 0;-6.690e-3 0 0], B=[0.9 -0.349 -0.15;0 14.155 0;0 -1.398 1.659], C=[1 0 0;0 1 0;6.34e-3 0 4.71e-3], D=[0 0 0;0 0 0;0.253 0.512 -0.014], x0=[108;66.65;428], u0=[0.34;0.69;0.436], y0=[108;66.65;0] Kp=[0.0582 -0.0196 0.9818;0.0018 0.0233 0.0002;-0.1158 -0.1516 5.8946], Ki=[0.0031 0.0023 0.0321 ;-0.0004 0.0093 -0.0001 ;-0.0004 0.0011 0.1925], Initial condition for Integrator=0

[Nabil]

Can anybody help me.

Answer 1

[Arkadiy Turevskiy]

Actually, your model should be stable. I created the model and it produces a stable response with the gains and parameters you specified. The response is slow, so to see it converge to steady state you need to change simulation time to 200. Could be that you are leaving simulation time at default 10, and then it looks like the response is unstable, when in reality it is stable - but just slow enough not to get to steady state in 10 secs.

[Nabil]

thank you for your answer, you are right after a while the response becomes stable,but I'm looking for a stable response of the first step and all along the simulation(the exact response) not just after a time, is that possible? Enclosed is my modelfor my thesis of Robust Controller Design for boiler system,I have to find the same response as in the article, I think something is missing or I incorrectly represented the model.